TfR1 interacts with the IKK complex and is involved in IKK–NF-κB signalling

NF-κB (nuclear factor κB) is the collective name of a family of transcription factors consisting of seven proteins, encoded by five genes: RelA, RelB, c-Rel, p105/p50 and p100/p52 [1]. All of these proteins are related via a highly conserved DNA-binding/dimerization domain, termed the Rel homology domain. Through this domain, NF-κB family members can form a variety of homo- and hetero-dimers to directly alter gene expression [1]. NF-κB-activation pathways can be classified as canonical, non-canonical and atypical regarding the mode and involvement of the IKK [IκB (inhibitor of NF-κB) kinase] complex [2].

IKK is a multimeric protein complex known to consist of at least three distinct subunits: two catalytic kinase subunits IKKα and IKKβ and a regulatory subunit IKKγ, also called NEMO (NF-κB essential modulator). IKKα and IKKβ are structurally related, but are functionally distinct, having different substrate specificities [3,4]. IKKγ, in contrast, has no catalytic activity, but instead is required for oligomerization and activation of the complex [3,4]. Despite recent findings of NF-κB-independent functions for IKK [5–8], the best understood role for these kinases is phosphorylation of IκB proteins and consequentially NF-κB activation [1].

IKK-mediated phosphorylation of IκBα generally triggers β-TrCP (β-transducin repeat-containing protein) binding, polyubiquitination and subsequently proteasomal degradation. The destruction of IκBα leads to the release of NF-κB dimers, translocation into the nucleus and binding to target gene promoters to either activate or repress their expression [1,4].

Early biochemical studies characterizing the IKKs revealed that they are part of a large molecular complex, of a size of approximately 900 kDa on size-exclusion chromatography analysis [9]. This suggests that other binding partners exist that could modulate the IKK complex activity and function. Indeed, proteins such as HSP90 (heat-shock protein 90) [10] and ELKS (glutamate-, leucine-, lysine- and serine-rich protein) [11] have been shown to interact with IKK and alter its function.

Given the importance of the IKK complex, we conducted an unbiased analysis for IKK-interacting proteins using quantitative MS. We have identified TfR1 (transferrin receptor 1), as a novel IKK-binding partner. Functionally, TfR1 is required for IKK complex activity, without altering IKK subunit levels. Importantly, depletion of TfR1 results in reduced NF-κB activation and increased apoptosis in response to TNFα (tumour necrosis factor α).

U2OS osteosarcoma and HEK (human embryonic kidney)-293 cells were grown in DMEM (Dulbecco's modified Eagle's medium) (Lonza) supplemented with 10% (v/v) fetal bovine serum (Gibco), 50 units/ml penicillin (Lonza) and 50 μg/ml streptomycin (Lonza) for no more than 30 passages. U2OS-NF-κB luciferase reporter cells were grown in DMEM supplemented with 10% (v/v) fetal bovine serum, 50 units/ml penicillin and 50 μg/ml streptomycin with 150 μg/ml Hygromycin B (Roche).

U2OS cells were transfected with 2 μg of NF-κB-reporter construct pGL4.32[luc2P/NF-κB-RE/Hygro] (Promega) using GeneJuice® (Merck Biosciences). After 48 h, cells were split and cultivated under 300 μg/ml Hygromycin B. Individual clones were picked and selected on the basis of their response to TNFα. Selected cell lines were grown in DMEM supplemented with 10% (v/v) fetal bovine serum, 50 units/ml penicillin and 50 μg/ml streptomycin with 150 μg/ml Hygromycin B.

The GFP (green fluorescent protein)–TfR1 expression construct was a gift from Professor Wolfhard Almers (Oregon Health and Sciences University, Portland, OR, U.S.A.). pCNA3-IKKβ-Flag and pCNA3-IKKα were gifts from Professor Ron Hay (Dundee University, Dundee, U.K.) pGL4.32[luc2P/NF-κB-RE/Hygro] was obtained from Promega. RSV (respiratory syncytial virus)-RelA was a gift from Professor Neil Perkins (Newcastle University, Newcastle upon Tyne, U.K.). pCMVTAG-IKKγ-Flag was a gift from Professor Jon Ashwell (National Cancer Institute, Bethesda, MD, U.S.A.) (Addgene plasmid 11970) [12].

siRNA duplex oligonucleotides were synthesized by MWG and transfected using Interferin (Polyplus) as per the manufacturer's instructions.

The siRNA sequences used were: Control, CAGUCGCGUUUGCGACUGG [13]; RelA, GCUGAUGUGCACCGACAAG [13]; TfR1_A, CCAAUACAGAGCAGACAUA; TfR1_B, ACUUGCUGUAGAUGAAGAA; TfR1_C, CUUCCAGACUAACAACAGA; and IKKγ, ACAGGAGGUGAUCGAUAAG [14].

Semi-quantitative RT–PCR and PCR was performed as described previously [14–16]. PCR products were resolved on 2% agarose gels and scanned using a PhosphoImager (FujiFilm FLA-5100) into TIFF format. Quantitative RT–PCR was performed using cDNA templates [cDNA synthesis was performed using Quantitect Reverse Transcription kit (Qiagen)] amplified using specific primer sets and the Stratagene Brilliant II SYBR® Green qPCR mix according to the manufacturer's instructions. Amplification and detection were performed using a Stratagene Mx3005P detection system. Sample values obtained with specific primer sets were normalized to β-actin primer set values.

The PCR primer sequences used were: actin, 5′-CTGGGAGTGGGTGGAGGC-3′ (forward) and 5′-TCAACTGGTCTCAAGTCAGTG-3′ (reverse); IAP2 (inhibitor of apoptosis protein 2), 5′-GTCAAATGTTGAAAAAGTGCCA-3′ (forward) and 5′-GGGAAGAGGAGAGAGAAAGAGC-3′ (reverse); p100, 5′-AGCCTGGTAGACACGTACCG-3′ (forward) and 5′-CCGTACGCACTGTCTTCCTT-3′ (reverse); and TfR1-Quantitect primer assay (Qiagen). ChIP (chromatin immunoprecipitation) primers used were: p100 κB, 5′-CACTCCGAGGAGGAGACACT-3′ (forward) and 5′-GGAGCAACCTTGGGATTTTC-3′ (reverse); and p100 control, 5′-TATGGGGGATTGAAGCAGAG-3′ (forward) and 5′-TGGGCCAAATGGAATACTGT-3′ (reverse).

Proteins were cross-linked with formaldehyde for 10 min. Glycine (0.125 mM) was added, and cells were washed with PBS. Cells were lysed with lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris/HCl, pH 8.1, 1 mM PMSF, 1 mg/ml leupeptin and 1 mg/ml aprotonin), followed by sonication and centrifugation. The supernatant was pre-cleared with sheared salmon sperm DNA and Protein G–Sepharose beads (Sigma). The supernatant was incubated with specific antibodies overnight, and then with Protein G–Sepharose beads for 1 h. After an extensive wash step, the complexes were eluted with buffer (100 mM NaHCO₃ and 1% SDS) and incubated with proteinase K. DNA was purified using QIAquick® PCR purification kit (Qiagen). PCR was performed on the purified immunoprecipitated DNA using specific primers.

Antibodies used were: anti-TfR1 (sc-51829, Santa Cruz Biotechnology), anti-phospho-RelA (Ser⁵³⁶) (3031, Cell Signaling Technology), anti-RelA (sc-372, Santa Cruz Biotechnology), anti-PCNA (proliferating-cell nuclear antigen) (P8825, Sigma), anti-β-actin (A5441, Sigma), anti-IκBα (sc-371, Santa Cruz Biotechnology; 4812, Cell Signaling Technology), anti-phospho-IκBα (Ser³²/Ser³⁶) (9246, Cell Signaling Technology), anti-phospho-IKKα/β (2681, Cell Signaling Technology), anti-IKKγ (sc-8330, Santa Cruz Biotechnology), anti-IKKβ (2678, Cell Signaling Technology), anti-IKKα (2682, Cell Signaling Technology; sc-7182, Santa Cruz Biotechnology), anti-GFP (Roche), anti-FLAG (Sigma), anti-IAP2 (3130, Cell Signaling Technology), anti-p100/p52 (05-361, Millipore) and anti-(cleaved PARP) [poly(ADP-ribose) polymerase] (9541, Cell Signaling Technology).

DFX mesylate (Sigma) was added for 2 h at a final concentration of 200 μM, ammonium iron(III) sulfate (Sigma) and ascorbate (Sigma) were also added for 2 h at final concentrations of 300 and 2 μM respectively.

Nuclear RelA blots were quantified using ImageJ (NIH), normalized with PCNA loading control and untreated control levels set to 1. All other conditions are compared with untreated non-targeting siRNA control levels.

Whole-cell protein extracts, nuclear extracts, EMSAs (electrophoretic mobility-shift assays) and luciferase assays were performed as described previously ([15] and references therein).

The IKK complex is central in the activation pathway of the transcription factor NF-κB [2]. As such, identification of a novel control mechanism would be of great interest for the understanding of the pathway, as well as for future therapeutic intervention. To unveil novel interacting proteins and potential novel regulators we employed the TAP (tandem affinity purification) technique followed by quantitative MS, utilizing SILAC (stable isotope labelling of amino acids in culture). We reconstituted IKKβ^−/− MEFs (mouse embryonic fibroblasts) with TAP–IKKβ or TAP alone (Supplementary Figure S1A at http://www.BiochemJ.org/bj/449/bj4490275add.htm). The TAP-tag consists of Protein A and a calmodulin-binding peptide separated by the recognition motif for the TEV (tobacco etch virus) protease. The functionality of the TAP–IKKβ-reconstituted MEFs was tested using NF-κB DNA binding following TNFα treatment. It was possible to see a rescue of IKK function in these cells, as measured by EMSA using an NF-κB consensus probe, to a level comparable with that of wild-type MEFs (Supplementary Figure S1B).

Having established the cellular system, the TAP–IKKβ- or TAP-reconstituted IKKβ^−/− MEFs were cultured in either ‘light’ (i.e. non-labelled) L-lysine and L-arginine or heavy-isotope-labelled amino acids L-[¹³C₆]lysine hydrochloride and L-[¹³C₆]arginine hydrochloride medium. Lysates were obtained from TAP–IKKβ cells (heavy) and TAP cells (light) mixed at an equal ratio, before being subjected to a sequential purification procedure as outlined in Supplementary Figure S1(C). The purified material was separated by SDS/PAGE (4–12% gels), and gel fragments were excised. A proteomic analysis was performed to identify and quantify the proteins that interact with TAP–IKKβ. Peptide ratios were quantified using MSQuant software to identify proteins that interact more with IKKβ. Comparison of the obtained peptide sequences with protein databases identified both IKKα and IKKγ, confirming that our purification technique pulls down known IKKβ-interacting proteins (Table 1). In addition, we also identified HSP90 as one of the IKKβ-associated proteins (Table 1). However, we did find a novel interaction partner for IKKβ: TfR1 (Table 1). We confirmed this interaction by Western blotting, using a specific antibody against TfR1 (Figure 1A).

To confirm that the interaction between TfR1 and IKK proteins is conserved in human cells, co-immunoprecipitation experiments between exogenous IKKα, IKKβ, IKKγ and TfR1 were performed. Our results demonstrate that TfR1 interacts with IKKα, IKKβ or IKKγ, but not with the negative controls (Figure 1B and Supplementary Figure S1D). To validate further our analysis, we performed co-immunoprecipitation experiments analysing endogenous proteins. This analysis confirmed that endogenous TfR1 interacts with all members of the IKK complex (Figure 1C). Finally, we tested whether under conditions of activation of the IKK complex, such as treatment with TNFα, TfR1 was still present in the complex. Our analysis revealed that TfR1 association with IKKβ or IKKγ does not change significantly with treatment with TNFα (Figure 1D).

To analyse the extent of association between TfR1 and the IKK complex, we analysed size-exclusion chromatography fractions for the presence of these proteins (Figure 2A). We found that endogenous TfR1 co-fractionates with the IKK complex subunits over several fractions, ranging from high-molecular-mass (1 MDa) to smaller complexes (Figure 2A). Since the bulk of TfR1 seems to co-fractionate mostly with IKKγ, we determined whether TfR1 was still able to associate with IKKα or IKKβ in the absence of IKKγ. Our analysis revealed that TfR1 remained associated with the catalytic IKKs, when IKKγ was depleted using siRNA (Figure 2B). These results indicate that TfR1 binds to the IKK complex in cells.

TfR1 is a single-pass type II membrane protein [17–19]. However, it does have a small intracellular portion, and it is internalized when bound to transferrin and subsequently recycled to the membrane again [17–19]. To test whether the intracellular portion of TfR1 was binding directly to IKK, we cloned the intracellular 64 residues of TfR1 into a bacterial expression construct. Protein fragments expressed in bacteria were purified and used in an in vitro binding assay with in vitro transcribed and translated IKKβ. Interestingly, we could not detect any specific binding of IKKβ to the small intracellular portion of TfR1 (Supplementary Figure S1E). These results suggest that the interaction is mediated by post-translational modifications on either of the proteins, requiring full-length TfR1, or it is indirect, via additional protein partners.

To address whether TfR1 influences IKK activity, we first sought to address whether IKK levels were altered by depleting TfR1 using RNAi (RNA interference). Relative to a non-targeting control, levels of the TfR1 protein were specifically decreased when U2OS cells were transfected with siRNAs against the TfR1 mRNA (Figure 3A). In contrast, TfR1 siRNA did not diminish the levels of the two IKK catalytic subunits IKKα and IKKβ or the regulatory subunit IKKγ.

The association between the regulatory subunit IKKγ, and the kinase subunits IKKα or IKKβ, is absolutely critical for the activity of the IKK complex [9]. To assess whether TfR1 was involved in IKK complex assembly, co-immunoprecipitation assays were used to investigate whether the integrity of the endogenous IKK complex is altered in the absence of TfR1. To this end, we depleted TfR1 using two different siRNAs, and immunoprecipitated IKKγ or IKKβ. Levels of associated IKKs were analysed by Western blotting (Figure 3B). Although expression of the IKK subunits is not altered by TfR1 depletion, association between the regulatory subunit IKKγ and the kinase subunit IKKβ is diminished in TfR1-compromised cells (Figure 3B). This is seen when either IKKβ or IKKγ are used to immunoprecitate the complex. Interestingly, IKKα interaction with IKKγ is still preserved, suggesting that TfR1 facilitates the IKKγ–IKKβ interaction, but has minimal effect on IKKγ–IKKα binding. These data indicate that TfR1 is required for the integrity of the full IKK complex.

The IKK complex responds to a number of stress stimuli, including cytokines such as TNFα [3]. Activation of the IKK complex can be assessed by phosphorylation in the T-loop, and by analysis of the phosphorylation status and levels of its substrate IκBα. To determine whether TfR1 is important for IKK activation following TNFα, TfR1 was depleted by siRNA and cells were treated with TNFα for different periods of time (Figure 4A). TNFα-induced IKK phosphorylation and activity were severely delayed and impaired when TfR1 was depleted (Figure 4A). As the IKK complex is the upstream kinase for the NF-κB family of transcription factors, we tested whether NF-κB transcriptional activity is altered in cells lacking TfR1. We created a U2OS cell line stably transfected with a NF-κB luciferase reporter construct. We initially tested NF-κB transcriptional activity in response to TNFα in a time-dependent manner (Supplementary Figure S2 at http://www.BiochemJ.org/bj/449/bj4490275add.htm). Luciferase activity was increased following 2 h of TNFα treatment and reached maximal levels following 6 h of treatment (Supplementary Figure S2). These results are in agreement with the NF-κB-activation pattern in most cells, which is cyclical in nature [20,21]. Using the NF-κB reporter cells, we depleted TfR1 using siRNA, and stimulated for 5 h with TNFα before harvest and luciferase assay. In cells depleted of TfR1, both basal and TNFα-induced NF-κB activity were reduced compared with control cells (Figure 4B). Indeed, the fold decrease in NF-κB activity observed in cells lacking TfR1 is comparable with that observed with depletion of RelA, the NF-κB family member that is most responsive to TNFα (Figure 4B).

Our results based on siRNA-mediated depletion indicate that TfR1 is required for IKK complex formation and activation in response to TNFα. To assess whether increased TfR1 levels increase IKK activity, we overexpressed TfR1 and assessed the NF-κB transcriptional response to TNFα stimulation (Figure 4C). Our analysis revealed that increased TfR1 levels result in a slight increase in NF-κB transcriptional activity, supporting further the hypothesis that TfR1 modulates IKK–NF-κB activity. In addition, we tested whether NF-κB activity could be modulated by iron levels. As such, we treated cells with either an iron chelator or increased iron, and analysed NF-κB luciferase activity following TNFα treatment. We observed that depletion of iron resulted in lower luciferase activity, whereas addition of extra iron to the cells potentiated NF-κB activity (Figure 4D). This could also be observed at the level of two endogenous NF-κB targets: IAP2 and p100 (Figure 4E).

Activation of the canonical NF-κB signalling pathway by TNFα requires the translocation of the RelA subunit from the cytoplasm to the nucleus where it binds target gene promoters and enhancers. To determine the role of TfR1 in this process, we transfected U2OS cells with a control or TfR1 siRNA oligonucleotides, treated with TNFα for different periods of time and isolated nuclear fractions. In control cells, we observed robust translocation of RelA from the cytoplasm to the nuclear compartment within 30 min of TNFα stimulation, whereas in cells depleted of TfR1, the levels of nuclear RelA were significantly decreased (Figure 5A). No change in total levels of RelA was observed; however, when TfR1 was depleted, there was a reduction in the levels of phosphorylated Ser⁵³⁶ on RelA (Figure 5A). These results are consistent with impairment of IKK function as this residue can be phosphorylated by IKK [22].

We investigated further the functional significance of this reduced RelA translocation/activation by analysing the effects of TfR1 depletion on the levels of mRNA from endogenous target genes by quantitative RT–PCR (Figure 5B). Induction of NF-κB target genes c-IAP2 (cellular IAP2) and p100 in response to stimulation with TNFα was compromised in cells depleted of TfR1 (Figure 5B). We also noted an induction of TfR1 mRNA by TNFα treatment at 4 h (Supplementary Figure S3A at http://www.BiochemJ.org/bj/449/bj4490275add.htm). This is in agreement with previous studies that have found that HIF-1 (hypoxia-inducible factor-1) and NF-κB can control TfR1 gene expression in response to inflammatory signals [23].

We further validated our results with Western blot analysis, where we observed that TNFα induction of IAP2 and p100 protein is impaired in the absence of TfR1 in U2OS cells (Figure 5C) and MDA-MB-231 cells (Supplementary Figure S3B). Importantly, in the absence of TfR1, RelA recruitment to the p100 promoter following TNFα treatment is visibly decreased (Figure 5D). Taken together, these results demonstrate that TfR1 is required for TNFα-mediated activation of endogenous IKK and NF-κB in the cell.

NF-κB signalling pathway in response to TNFα is to transactivate anti-apoptotic genes, in order to protect the cell against programmed cell death [1]. As we have demonstrated that depletion of TfR1 interferes with the IKK–NF-κB signalling pathway, we sought to determine whether loss of TfR1 could interfere with the anti-apoptotic functions associated with the IKK–NF-κB axis. We depleted U2OS cells of TfR1 and then treated them with increasing concentrations of TNFα for 24 h before lysis. Western blot analysis was performed to determine the levels of the known apoptosis marker cleaved PARP. In control cells, TNFα does not lead to significant levels of cell death owing to NF-κB anti-apoptotic function (Figure 6A). However, in cells lacking TfR1, TNFα-induced apoptosis is readily observed even with low concentrations of TNFα, as these cells display much higher levels of cleaved PARP when compared with control cells (Figure 6A).

To rule out the possibility that the effects of TfR1 depletion on TNFα-induced apoptosis are not dependent on modulation of IKK–NF-κB, we performed rescue experiments. Cells were depleted of TfR1, but, in addition, we expressed control or a plasmid for the NF-κB subunit RelA, treated with TNFα and analysed the levels of cleaved PARP. Expression of exogenous RelA visibly reduced the levels of cleaved PARP observed in cells depleted of TfR1 and treated with TNFα (Figure 6B). The results confirm the role of TfR1 in the modulation of IKK–NF-κB function in cells.

The IKK complex controls the activity of a very important family of transcription factors called NF-κB. Given the role of NF-κB in the immune system and in pathological conditions, the IKK complex has become an attractive drug target for pharmaceutical intervention [1]. As such, increasing our understanding of how the IKK complex is regulated would be of great interest. For this purpose, we performed a quantitative analysis for IKK-interacting proteins using quite stringent purification procedures and identified TfR1 as a novel binding partner for the IKK subunits in cells (Figures 1 and 2). Functionally, we found that TfR1 is required for IKK complex formation and activity (Figures 3–6). As such, depletion of TfR1 results in lower IKK and NF-κB activation and increased sensitivity to apoptosis following TNFα (Figures 4–6).

TfR1 is a type II transmembrane protein that is the primary cell-surface receptor that regulates uptake of iron-bound transferrin by receptor-mediated endocytosis [18,19]. It plays an essential role in iron homoeostasis [18,19].

As iron is essential for cell-cycle progression, TfR1 expression is up-regulated in dividing cells, and has been found to be elevated various types of cancer [17,24]. Given these observations, it has become an attractive target for therapy, with the use of inhibiting antibodies and TfR1–toxin fusion molecules [17,25].

TfR1 has been implicated previously, although indirectly, in NF-κB signalling. Gambogic acid, a ligand for TfR1, has been shown to block TNFα-induced NF-κB signalling. This may be through disruption of IKKβ signalling [26], or perhaps covalent modification of the IKKβ subunit of IKK [27], but the exact mechanism remains unclear. In the present study, we have shown that TfR1 is a novel IKK-interacting protein. We have demonstrated that TfR1 can regulate NF-κB by altering the activity of the IKK complex. Loss of TfR1 alters the activity and integrity of the complex, which subsequently compromises the expression of NF-κB-target genes. These functional results are consistent with previous results using gambogic acid [26,28,29]. Although we could not find an in vitro interaction between the cytoplasmic tail of TfR1 and IKK, these proteins do interact in cells, and it suggests that TfR1 interacts with the IKK complex through a different domain, or via an intermediate protein. Interestingly, TfR1 was identified as an interacting protein of IKKϵ [30], a related IKK that has also been shown recently to control canonical IKK activity [31,32].

Studies using gambogic acid have shown that HSP90 could be involved in the control of IKK activity [33]. Zhang et al. [33] have shown that HSP90 is inhibited by gambogic acid, leading to reduced expression of HSP90 and IKKs. We did not find any changes in the levels of IKK subunits in the absence of TfR1 (Figure 3A). However, it is still possible that TfR1–IKK complex assembly could be mediated by changes in HSP90.

The control of IKK by TfR1 raises the interesting possibility that NF-κB can respond to iron changes in the cells. Previous studies have shown that iron chelation leads to NF-κB inhibition [34,35]; however, the mechanism behind these observations is unknown. Similarly, some studies have shown NF-κB activation in response to iron increases [36]. In the present study, we have shown that both depletion and overexpression of TfR1 changes NF-κB activity in cells, suggesting that TfR1 could mediate iron modulation of NF-κB activity. We have also observed increased NF-κB activity when iron was added to the cells or when TfR1 was overexpressed, demonstrating that NF-κB does sense changes in iron levels in the cell. Of interest, NF-κB can also induce TfR1 expression via regulation of HIF-1α levels [13,23,37], establishing another connection between these molecules. Given our results, it would be interesting to test whether TfR1 depletion alters the levels of HIF-1α following TNFα or other inflammatory signals.

The importance of TfR1-mediated control of IKK–NF-κB activity is demonstrated by the sensitization of cells to TNFα-induced apoptosis (Figure 6). TNFα activates both apoptotic [FADD (Fas-associated death domain)/caspase 8] and anti-apoptotic (NF-κB) pathways in the cell. As such, IKKβ-, IKKγ- or RelA-deficient cells are extremely sensitive to TNFα-induced apoptosis, whereas the normal counterparts are resistant [38–41]. This is also true for the knockout mice for these proteins that are embryonic lethal owing to liver apoptosis, which can be rescued by knockout of the TNFα receptor [39–41]. Our results show that, in the absence of TfR1, IKK–NF-κB is not able to prevent apoptosis following TNFα, and, importantly, this apoptosis can be rescued by increasing the levels of RelA in cells. This indicates that the effects seen on apoptosis following TfR1 depletion are NF-κB-dependent and not due to a different role of TfR1 in the cell.

As NF-κB has mostly pro-survival functions in advanced tumour stages [42], our results suggest that TfR1 would be a good therapeutic target for cancer types with elevated IKK–NF-κB activity.

In summary, we have demonstrated that TfR1 binds to IKK and controls complex assembly and activity, which has physiological implications for the levels of NF-κB activation following cytokine stimulation. It also provides a mechanism for effects seen using TfR1 inhibitors and modulators of iron levels in cells.

ChIP

chromatin immunoprecipitation

DFX

desferoxamine

DMEM

Dulbecco’s modified Eagle’s medium

EMSA

electrophoretic mobility-shift assay

GFP

green fluorescent protein

HEK

human embryonic kidney

HIF-1

hypoxia-inducible factor-1

HSP90

heat-shock protein 90

IAP2

inhibitor of apoptosis protein 2

c-IAP2

cellular IAP2

IκB

inhibitor of nuclear factor κB

IKK

IκB kinase

MEF

mouse embryonic fibroblast

NF-κB

nuclear factor κB

PARP

poly(ADP-ribose) polymerase

PCNA

proliferating-cell nuclear antigen

RT

reverse transcription

siRNA

small interfering RNA

TAP

tandem affinity purification

TfR1

transferrin receptor 1

TNFα

tumour necrosis factor α

Niall Kenneth, Sharon Mudie, Sanne Naron and Sonia Rocha performed experiments. Niall Kenneth, Sharon Mudie and Sonia Rocha analysed the data. Niall Kenneth and Sonia Rocha initiated the study. All authors contributed to writing the paper.

We thank Professor Ron Hay, Professor Wolfhard Almers, Professor Jon Ashwell and Professor Neil Perkins for providing valuable reagents.

This study was supported by funding from Tenovus Scotland, Association for International Cancer Research (AICR) and Cancer Research UK. S.N. is funded by a Biotechnology and Biological Sciences Research Council studentship.