Cell Survival from Chemotherapy Depends on NF-κB Transcriptional Up-Regulation of Coenzyme Q Biosynthesis

Background Coenzyme Q (CoQ) is a lipophilic antioxidant that is synthesized by a mitochondrial complex integrated by at least ten nuclear encoded COQ gene products. CoQ increases cell survival under different stress conditions, including mitochondrial DNA (mtDNA) depletion and treatment with cancer drugs such as camptothecin (CPT). We have previously demonstrated that CPT induces CoQ biosynthesis in mammal cells. Methodology/Principal Findings CPT activates NF-κB that binds specifically to two κB binding sites present in the 5′-flanking region of the COQ7 gene. This binding is functional and induces both the COQ7 expression and CoQ biosynthesis. The inhibition of NF-κB activation increases cell death and decreases both, CoQ levels and COQ7 expression induced by CPT. In addition, using a cell line expressing very low of NF-κB, we demonstrate that CPT was incapable of enhancing enhance both CoQ biosynthesis and COQ7 expression in these cells. Conclusions/Significance We demonstrate here, for the first time, that a transcriptional mechanism mediated by NF-κB regulates CoQ biosynthesis. This finding contributes new data for the understanding of the regulation of the CoQ biosynthesis pathway.


Introduction
Coenzyme Q (CoQ) is a small lipophilic molecule that transports electrons from mitochondrial respiratory chain complexes I and II, to complex III [1]. In addition, CoQ functions as a cofactor for uncoupling proteins [2] and other mitochondrial dehydrogenases [1]. CoQ mainly acts as an antioxidant and can prevent cell death under certain stress conditions, particularly in mitochondria-DNA depleted cells [3,4]. CoQ also regulates the extracellulary induced ceramide-dependent apoptotic pathway [5]. CoQ is composed of a benzoquinone ring and a polyisoprenoid chain, derived from tyrosine and mevalonate, respectively. Its biosynthesis depends on a pathway that involves at least ten genes (COQ genes). Among them, COQ7 is proposed to encode for a key regulatory component of a multisubunit enzyme complex [6]. However, there is no information about the precise regulation of CoQ biosynthesis pathway except that peroxisome proliferator-activated receptor alpha (PPARa) is involved [7].
We have previously shown that Campothecin (CPT) treatment increases CoQ biosynthesis rate and reported that CPT in mammals up-regulates COQ7 [8]. Thus, the stimulation by CPT is a useful tool for deciphering transcriptional mechanisms of the regulation of the CoQ biosynthetic pathway in mammals through up-regulation of COQ7 gene. Camptothecin (CPT) is a cytotoxic drug widely used in cancer therapy. It is known that the main target for camptothecin is the nuclear topoisomerase I (TOP1) [9][10][11]. Double-strand DNA breaks derived from the inhibition of nuclear TOP1 are considered the main cause of apoptosis induction by CPT [11]. CPT also induces an increase of reactive oxygen species (ROS) in different cancer cell lines including H460 cells [12,13,14,15,16,8].
There are recent reports supporting the role of oxidative stress in the induction of apoptosis by CPT and its derivatives [15,17]. In response to topotecan, a CPT water-soluble derivative, cells activate their antioxidant defense mechanisms and a number of antioxidant enzyme activities, such as catalase, manganese-dependent superoxide dismutase (MnSOD), and glutathione peroxidase [16]. Also, the addition of catalase was able to protect cells from CPT induced apoptosis in HL-60 leukemia cells [12]. Furthermore, catalase administration to U-937 promonocytic cells also attenuated apoptosis induction by CPT and other cytotoxic drugs [18]).
NF-kB is a redox-sensitive transcription factor, which regulates antioxidant enzymes such as MnSOD encoded by the SOD2 gene. NF-kB is also activated by CPT in several cell types [19][20][21][22]. In fact, NF-kB activation is frequently abrogated by antioxidants [23,24]. NF-kB has been shown to play a key role in the regulation of cell death, either as inducer or, more often, as blocker of apoptosis, depending on the cellular type and the insult [25,26].
Thus, we have proposed that NF-kB could be one of the mediators of the cellular effects by CPT through the activation of the CoQ biosynthesis pathway.

CoQ biosynthesis is dependent on NF-kB
Oxidative stress rises as an important activator of NF-kB that can be abrogated by antioxidants [23,24]. We have previously demonstrated that in H460 cells, which were blocked in the CoQ biosynthesis pathway, exhibited a increased sensitivity to production of ROS and cell death induced by CPT [8].
H460 cells treated with 10 mM CPT for 24 hours were fixed and probed with p65 antibody to confirm that the NF-kB system is active in these cells. Immunofluorescence experiments showed that the transcription factor translocated into the nucleus in response to CPT ( figure 1 A).
In order to test whether NF-kB elicits a survival response in H460 cells and if the induction of CoQ biosynthesis in CPT treated cells is dependent on NF-kB activation, cell viability and cellular CoQ levels were measured in the presence of Bay 11-7085, a specific inhibitor of NF-kB [27]. The inhibition of NF-kB increased cell death and reduced CoQ levels induced by CPT ( figure 1 B and C). In parallel, we observed that the increase of COQ7 mRNA measured by real time PCR in CPT treatments was significantly abolished by Bay 11-7085 (figure 1 D), suggesting a dependence of COQ7 gene to NF-kB.
It has previously been reported that NF-kB system is downregulated in DU145 RC0.1, a cell line resistant to CPT [28]. We used parental DU145 and RC0.1 cells to evaluate protein levels of the NF-kB system components p65 and IkBa and observed a very low content in RCO.1 compared to parental cells (figure 2 A).
In RC0.1 cells, CPT treatment induced a higher production of ROS compared to parental DU145 and H460 cells (figure 2 B). Since mitochondrial ROS generation stimulated by CPT is responsible for the induction of CoQ biosynthesis, CoQ levels and COQ7 mRNA were measured in both parental DU145 and RC0.1 cell lines after 10 mM CPT treatment for 24 h. The parental DU145 cell line showed an increase in both CoQ levels and COQ7 messenger in response to CPT. However, the RC0.1 cells showed only a trend to increase CoQ levels and COQ7 messenger figure 2 C and D). Since the NF-kB system is down regulated in RC0.1 cells, this finding would confirm the requirement of NF-kB to CPT induced CoQ biosynthesis. NF-kB specifically binds to COQ7 kB sites Transcription element search software (TESS) sequence analysis [29,30] of 4 Kb of the 59-flanking region, exon 1 and the beginning of the first intron of the human COQ7 gene revealed two potential binding sites for NF-kB, three for Sp1, as well as one for RXR ( figure 3). The same analysis revealed other putative binding sites but we decided to focus on those potentially related to the oxidative stress response, particularly NF-kB because it is a redox-sensitive factor that has been shown to be activated by CPT [19]. One of the putative NF-kB binding sites, kB1 site, was found in the promoter region of the gene (position 2360 to 2350) whereas the second one, kB2, was identified at the beginning of the first intron of the gene (position +120 to +131).
Electrophoretic mobility shift assays were performed to evaluate the binding of the NF-kB transcription factor to the hypothetical binding sites found in COQ7 gene. Nuclear extract from CPT treated cells were incubated with digoxigenin-labeled oligonucleotides encompassing kB1 or kB2 putative binding sites. The incubation with kB1 probe generated a band shift in response to CPT treatment (figure 4 A). The intensity increase of the shifted band achieved significance after 1 hour of treatment. Competition assays with an excess of either unlabeled normal or nonspecific oligonucleotides confirmed the specificity of the binding (figure 4 B). In addition, competition assays using antibodies against diverse subunits of NF-kB were also performed. Anti p50 impaired the formation of the protein-DNA complex. However, it failed to react with anti p65 and anti p52 sera (figure 4 C).
We assayed the time course binding of NF-kB to kB2 site (figure 5 A). Incubation of the kB2 probe with nuclear extracts of CPT treated cells resulted in a band shift that reached a maximum intensity at 6 hours of treatment. In order to demonstrate that the in vitro NF-kB binding to kB2 was specific and that the composition of the heterodimer was p65/p50 we ran competition (figure 5 B) and super-shift assays (figure 5 C).
Our results strongly support the in vitro specific binding of NF-kB to both sites in the COQ7 sequence in response to CPT.

COQ7 kB binding sites are functional
To evaluate the physiological importance of the COQ7 kB binding sites, firefly luciferase reporter constructs containing several regions of the COQ7 59-flanking sequence (figure 6 A) were transfected into HeLa cells. We have previously demonstrated that CPT also induces cell death and stimulates CoQ biosynthesis in HeLa cells [8]. Transfected cells were treated with 10 mM CPT and their luciferase activity was measured. Results represent firefly luciferase activity normalized on the basis of the constitutive b-gal activity of a control vector that was cotransfected. Three hours of CPT treatment induced a significant increase in the reporter activity when pGL3-2 construct was assayed. Reporter assays performed with pGL3-1 construct, with a deletion of 1150 bp in 59, did not show response to CPT (figure 6 B). These data then suggest that it is necessary a large region consisting of at least 2770 bp are required for COQ7 to respond to CPT. The presence of the NF-kB inhibitor Bay 11-7085 during the CPT treatment of the transfected cells completely abolished the induction of the reporter gene expression (figure 6 B).
To further understand the system, deletions of seven bp inside the kB binding sites were made to generate the pGL3-2kB1, pGL3-2kB2 and pGL3-2kB1+kB2 plasmids. None of the constructs responded to CPT (figure 6 C), suggesting that both binding sites are necessary for the transcriptional induction of COQ7 in response to CPT.
To determine which IkB isoform was involved in the NF-kB activation by CPT, cytosolic levels of both isoforms were determined by western blot. Both IkBa and IkBb were degraded following the CPT treatment, which was prevented by Bay 11-7085. On the other hand, Bay 11-7085 also inhibited the nuclear translocation of NF-kB subunit p50 (figure 6 D).

Discussion
The lipidic antioxidant CoQ is considered a central component of the antioxidant defense, protecting cells from membrane peroxidations and regenerating the reduced forms of exogenous antioxidants [31]. CoQ has also been proposed to prevent apoptosis derived from oxidative stress induced by different stimuli [32,33].
The details about the CoQ biosynthesis pathway are unfolding. There are at least ten genes (COQ) involved in a complex biosynthetic pathway [34,6]. However, there is still little knowledge about its regulation [1], except that CoQ biosynthesis is activated in mouse liver by the nuclear regulator PPARa [35]. As the number of identifiable human pathologies that are associated with a primary CoQ deficiency increases, there is a need for the full understanding of its biosynthetic pathway, from the proteins participating in the enzymatic process, to the regulatory mechanisms [36].
We have previously shown that CPT induces the up-regulation of COQ7 [8], which is considered to have both a regulatory and kinetic role in CoQ biosynthesis [37,6]. Thus, we have focused on the regulatory mechanisms of COQ7 expression as a marker of CoQ biosynthetic pathway transcriptional regulation. Searching for putative binding sequences for known transcription factors we have identified two hypothetical binding sites for NF-kB through the analysis of the COQ7 59 sequence using TESS program based on TRANSFACT database v6.0 [29,30].
NF-kB activation has been observed in diverse physiological processes including immune response regulation, inflammation and development [25]. It is also active in some tumors, during chemotherapy response and under oxidative stress [38][39][40]. In fact, NF-kB activation by many agents can be abrogated by antioxidants [23,24]. Moreover, it is known that the antioxidant enzyme MnSOD is transcriptionally activated through NF-kB in response to oxidative stress [41,42,21,22]. NF-kB is generally accepted to be activated by the nuclear DNA damage exerted by CPT [19]. However, DNA damage might not be the only process by which CPT triggers NF-kB, as the antioxidant pyrrolidine dithiocarbamate (PDTC) is very effective in abolishing NF-kB activation by the drug [23].
There is no information about NF-kB involvement in CoQ biosynthesis regulation, which positively responds to oxidative stress conditions such as vitamin E withdrawal and aging [43][44][45]. Our results demonstrate that NF-kB is able to specifically bind to both COQ7 kB sites in response to CPT. Super-shift assays indicated that the dimer that recognizes the kB2 site is composed of both p65 and p50 subunits and the homodimer p50/p50 probably binds to kB1. It is important to note that NF-kB activation by CPT has been extensively studied. However, assays have generally been done using consensus sequences corresponding to either immunoglobulin heavy chain gene [19] or the HIV enhancer [23], but we have used here a consensus sequence specific for COQ7 gene. Thus, although it has been defined p65/ p50 as the only complex involved in CPT-mediated NF-kB activation [23], we can not rule out the possibility that the COQ7 kB2 site is recognized by a p50/p50 homodimer.
The specific binding of NF-kB to a DNA probe does not demonstrate its functionality in vivo but luciferase assays demonstrated that COQ7 was transcriptionally activated by CPT. The different constructs assayed have shown that it was necessary to  have at least 2150 bp of the 59 flanking region for COQ7 to respond to the drug. A construct containing 1000 bp of the 59 flanking region (pGL3-1) was not able to respond to CPT. TESS analysis of the sequence of this region shows the presence of putative binding sites for Sp1 and RXR transcription factors. Interestingly, NF-kB and Sp1 have been found to cooperatively bind to DNA in different promoter systems [46][47][48]. Moreover, interaction between NF-kB and Sp1 has been also found in the Firefly luciferase activities expressed in HeLa cells transfected with two different reporter constructs (pGL3-1 and pGL3-2) under different treatments. Luciferase activity was normalized on the basis of the b-gal activity constitutively expressed by the co-transfected plasmid pC110. Results are the mean6SD of at least three independent transfection experiments. BAY indicates the presence of the NF-kB inhibitor Bay 11-7085. * P,0.05, ** P,0.001, a P = 0.05 between the treatment and its control. C. Deletion analysis of the COQ7 putative kB sites. The kB sites were partially deleted as indicated in blue. CPT responsiveness of these constructs was assayed as described before. Results are expressed as the relative CPT/control response in transfected cells. Results are the mean6SD of three independent transfection experiments. D. Western blot analyzing the presence of different NF-kB system elements on the cytosolic or nuclear compartments. Punceau staining of the membrane used for the Western blot is presented as loading control. doi:10.1371/journal.pone.0005301.g006 promoter of the oxidative stress responsive gene SOD2 [22]. On the other hand, RXR has been shown to be required for CoQ biosynthesis and its induction by cold exposure in mice [35].
Both basal binding and CPT-activated COQ7 mRNA transcription promoter are inhibited by the NF-kB inhibitor Bay 11-7085, which targets the NF-kB activating IkB kinase complex (IKK) [27], supporting the NF-kB participation in the transcriptional regulation of the gene. Inhibition of NF-kB by Bay-117085 not only avoided the CPT-dependent increase of COQ7 mRNA, but also the increase in CoQ levels. Additionally, viability assays have shown that the inhibition of the transcription factor sensitizes cells to die by CPT. All these results further support the hypothesis that CPT triggers a NF-kB survival response involving the antioxidant CoQ.
The treatment of cells with chemotherapeutic compounds such as CPT induces cell death but some cell types can survive via activation of antioxidant pathways [49], including the increase of CoQ biosynthesis [8]. Here we have shown that CPT increases ROS production. This situation activates NF-kB that, among other factors, induces the expression of COQ7 gene by transcriptional regulation, which, in turn, increases CoQ biosynthesis. These findings contribute to the understanding of the proteins that participate in the enzymatic process of CoQ biosynthesis, which will help further our understanding of the CoQ 10 biosynthetic pathway and its involvement in CoQ deficiency syndrome in humans.

Cell viability measurement
Dead cells become propidium iodide (PI) permeable as a consequence of the lost of plasma membrane integrity. Thus, cell viability was determined by PI (10 mg/ml) staining and flow cytometry analysis.

Measurement of CoQ levels
Lipid extraction from cells samples was performed as described previously [8]. CoQ6 or CoQ9 were used as internal standard. Cell samples were lysed with 1% SDS and vortexed for 1 min. A mixture ethanol:isopropanol (95:5) was added and the samples were vortexed for 1 min. To recover CoQ, 5 ml of hexane was added and the samples were centrifuged at 10006g for 5 min at 4uC. The upper phases from three extractions were recovered and dried on a rotatory evaporator. Lipid extract were suspended in 1 ml of ethanol, dried in a speed-vac and kept at 220uC until analysis. Samples were suspended in the suitable volume of ethanol prior to HPLC injection. Lipid components were separated by a Beckmann 166-126 HPLC system equipped with a 15-cm Kromasil C-18 column in a column oven set to 40 uC, with a flow rate of 1 ml/min and a mobile phase containing 65:35 methanol/n-propanol and 1.42 mM lithium perchlorate. CoQ levels were analyzed with an ultraviolet (System Gold 168), electrochemical (Coulochem III ESA) and radioactivity detector (Radioflow Detector LB 509, Berthod Technologies) when necessary.

ROS measurement by flow cytometry
Free radical measurement was achieved by H 2 DCF-DA or CM-H 2 DCF-DA (Molecular Probes). These compounds are cellpermeant indicators for reactive oxygen species that are nonfluorescent until removal of the acetate groups by intracellular esterases and oxidation occurs within the cell. H460 cells were seeded in 35-mm dishes and treated when confluence was reached. Cells were incubated with 10 mM CM-H 2 DCFDA or H 2 DCF-DA for the last 30 min of treatment, washed and detached with trypsin. Flow cytometric analysis was performed with Epics H XL cytometer (Coulter) after trypsin removal. Population was selected using FS and SS and the fluorescence was measured by FL1 detector (525620 nm). H 2 O 2 was used as a positive control for detection of cellular free radicals.

Immunoblotting
Mitochondrial, nuclear or cytosolic fractions were resolved in a SDS-polyacrilamide gel electrophoresis (SDS-PAGE) (with variable acrilamide percentage according to the molecular weight of the protein of interest) and transferred to nitrocellulose membranes in a semi-dry transfer system (Trans-Blot, Biorad). After verification of equal loading using Ponceau S, membranes were blocked with 5% non-fat milk in TBS buffer (1 h room temperature or over night at 4uC) and stained with the appropriate primary antibodies (2 h at room temperature or over night at 4uC). Anti IkBa, IkBb p50 and actin were diluted 1:1000. Incubation with anti rabbit (1:10000) or anti mouse (1:5000) HRP-conjugated secondary antibodies was performed during 2 hours at room temperature. Immunolabeled proteins were detected by membrane exposure to X-ray film after incubation in enhanced chemiluminiscence reagent (Immun-Star HRP Substrate Kit, Biorad).

Real Time PCR
Relative expression levels were determined by real-time PCR. Cells were seeded in 35-mm plates and treated appropriately. Floating-dead cells were discarded and monolayer was washed with cold PBS. Total RNA from cells cultures was extracted with the TriPure Isolation Reagent (Roche). RNA preparation was treated for DNA removal with deoxiribonuclease I (Sigma) and cDNA was obtained from 1 mg of RNA by using the iScript cDNA Synthesis Kit (Biorad). Real Time PCR was performed in a MyiQ TM Syngle Color Real Time PCR Detection System (Biorad) coupled to a Biorad conventional thermocycler. Primers for COQ7 and the housekeeping gene were designed with the Beacon Designer 4 software. Amplification was carried on with iQ SYBR Green supermix (Biorad) with the following thermal conditions: 30 s at 95uC and 35 cycles of 30 s at 94uC, 30 s at 60uC and 30 s at 72uC. All the results were normalized to the levels of actin mRNA. At least three independent experiments were performed and the results averaged.

Electrophoretic mobility shift assay (EMSA)
Synthetic oligonucleotides encompassing the putative kB sites of the promoter region of the human COQ7 gene (kB1F-TAAAG-CAGGAAATACCGTGCCT; kB1R-AGGCACGGTATTTC-CTGCTTTA; kB2F-AGCTAGGGAATTTTCGCTTGA; kB2R-TCAAGCGAAAATTCCCTAGCTC) were purchased from MWG Biotech AG (Germany) as single stranded oligonucleotides. Digoxigenin labelling of the oligonucleotides was performed following the DIG Gel Shift 2nd Generation Kit protocol (Roche). Briefly, complementary forward and reverse oligonucleotides were denatured 2 min at 95uC and placed at room temperature for 30 min to generate double stranded oligonucleotides. A reaction was set up in a final volume of 20 ml in 1X reaction buffer (0.2M potassium cacodylate, 25 mM Tris-HCl, 0.25 mg/ml BSA (pH 6.6) at 25uC) containing 4 pmol of double stranded oligonucleotide, 5 mM CoCl 2 , 0.05 mM digoxigenin-ddUTP and 20 U of terminal transferase and incubated 15 min at 37uC. The reaction was stopped with 2 ml of 0.2 M EDTA (pH 8.0). Resultant labelled probe was diluted to a final concentration of 16 fmol/ml with TNE (Tris-HCl 10 mM; NaCl 100 mM; EDTA 1 mM).
Preparation of nuclear extract was performed by an adaptation of the protocol described by Schreiber and co-workers [50]. Briefly, the day before the treatment H460 cells were seeded at a density of 8 10 4 cells/cm 2 . After the treatment, cells were directly lysed on the flask by the addition of an hypoosmotic buffer (10 mM Hepes, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.625% NP-40, 1 mM DTT, 0.5 mM PMSF and protease inhibitor cocktail (Sigma)). Cell lysate was incubated for 5 min on ice, vigorously vortexed during 10 s and centrifuged at maximum speed for 30 s. Supernatant was reserved as cytoplasmic extract and stored at 280uC. The pellet was resuspended in an hyperosmotic buffer (20 mM Hepes, 0.4M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF and protease inhibitor cocktail (Sigma)) and incubated on ice during 15 min in agitation. Extracts were centrifuged at 16000 g 30 min at 4uC and supernatants were recovered as nuclear protein extracts without DNA.
Electrophoretic mobility shift assays (EMSAs) were performed following the DIG Gel Shift 2nd Generation kit (Roche) indications: 20 mg of nuclear extract were incubated in 1X Binding Buffer (20 mM Hepes pH 7.6, 1 mM EDTA, 10 mM (NH 4 )SO 4 , 1 mM DTT, 0.2% Tween-20, 30 mM KCl), 1 mg poly [d(I-C)], 0.1 mg poly Llysine and 64 fmol of DIG-labelled probe. Competition assays were performed in the presence of a 10X excess of cold specific or non specific (Oct2A: Forward strand GTACGGAGTATCCAG-CTCCGTAGCATGCAAATCCTCTGG; Reverse strand CCT-CATAGGTCGAGGCATCGTACGTTTAGGAGACCAGCT) probe. For supershift assays 0.4 mg of rabbit polyclonal antibody (p65, p50 or p52) was pre-incubated with the nuclear extracts for 30 min on ice before the binding incubation. Samples were loaded in a 4-8% native acrilamide: bisarilamide (30:1) gel (0.5X TBE buffer) and electrophoresed at 150 V in 0.5X TBE buffer. Electrotransference to a nitrocellulose membrane (Hybond TM -N+, Amersham) was performed with a Trans-Blot Semi-Dry system (Biorad) following the manufacturer's instructions. Samples were fixed to the membrane by UV-crosslinking (70 mJ/cm 2 ). Dig-labelled probes were detected with anti-digoxigenin and visualized with the chemiluminescent substrate CSPD from the DIG Gel Shift 2nd Generation Kit (Roche) and the membrane exposure to an X-ray film.
Cloning of COQ7 promoter region and generation of luciferase reporter constructs Inserts in pGL3-1 and pGL3-2 vectors were generated by PCR (special thermal conditions: 6 cycles with an annealing temperature of 42uC, followed by a Touch Down-PCR starting at an annealing temperature of 62uC and finishing at 42uC) from the BAC clone RP11-626G11 (AC099518; BACPAC Resource Centre, Children's Hospital Oakland Research Institute). COQ7 promoter region was first cloned into pGEM-T Easy Vector (Promega) using primers containing the restriction enzyme sites for SacI (59) and HindIII (39): hCQ7pSacIF (GTAGAGCTCTCCAAGGGTGTAA), CQ7pHin-dIIIR (GTTAAGCTTGTCCTGTTCACAG), for pGL3-1, and hCQ7pSacI3 (GTAGAGCTCACAGAGGGAGG) and CQ7pHindIIIR for pGL3-2. Primers were designed to amplify the 59 flanking region of the COQ7 gene (1000 pb for pGL3-1 and 2150 for pGL3-2), the 59 UTR region (60 bp), the complete first exon (73 bp) and, partially, the first intron of the gene (487 bp). Once verified by sequencing, fragments were subcloned into the pGL3basic vector (a generous gift from Dr. Blesa, Centro de Investigación Príncipe Felipe, Valencia, Spain). Simple and double kB sites deletions for the pGL3-2DkB1, pGL3-2DkB2 and pGL3-2DkB1+2 vectors were carried out by using the Quikchange II XL Site-Directed Mutagenesis Kit (Stratagene) with the following primers pairs: for kB1 site, CCTTCCACGTAGGTAAAGCAGC-GTGCCTTGTTAATAAGTAAT and ATTACTTATTAA-CAAGGCACGCTGCTTTACCTACGTGGAAGG; for kB2 site CGGTCTAGCGAGCTAGGCGCTTGAGGTTTGGGTC and ACCCAAACCTCAAGCGCCTAGCTCGCTAGACCG.

Reporter gene assays
Endotoxin-free plasmids were obtained by Qiagen EndoFreeH Plasmid Maxi Kit, to avoid unspecific activation of NF-kB. HeLa cells were transfected with FuGENE 6 (Roche) following the manufacturer's instructions. Briefly, the day before transfection, 2.9610 5 cells were plated in 12 well dishes (DMEM 10% serum without antibiotics). One hour before transfection, media was changed. For each single transfection 2.2 ml of FuGENE reagent were diluted in 50 ml of non supplemented OPTIMEM media (Gibco) and incubated for 10 minutes at room temperature. 0.74 mg of total DNA was added to the FuGENE diluted transfection mixture and further incubated for 30 minutes at room temperature. Then, the transfection reagent/DNA complex was added to the cells. Cells were co-transfected with pGL3-1 or pGL3-2 (0.37 mg) and pCH110 (0.37 mg), a plasmid containing the bgalactosidase gene regulated by a constitutive promoter (a generous gift from Dr. Carrión, Universidad Pablo de Olavide, Sevilla, Spain). The transfection proceeded for 48 hours before cells were appropriately treated. After that, death cells were discarded and monolayer was washed with cold PBS and lysed in 100 ml of Luciferase Lysis Buffer (Promega). Lysates were vigorously vortexed for 10 s and centrifuged 12000 g during 15 min. Aliquots were stored at 280uC until used. b-galactosidase activity was quantified in the same extracts used for luciferase activity measurement. 30 ml of lysate sample was mixed with 20 ml of H 2 O and 50 ml of reaction buffer (200 mM Na 2 HPO 4 pH 7.3, 2 mM MgCl 2 , 100 mMb-mercaptoethanol, 1.33 mg/ml ONPG [onitrophenyl-b-D-galactopyranoside]). Samples were incubated for 45 min/ 1 h at 37uC and b-galactosidase activity was spectrophotometrically measured at 405 nm in a Sunrise plate reader (Tecan Austria GmbH) coupled to the Magellan software. Luciferase activity was measured by a manual luminometer Luminoskan TL Plus (Thermo LabSystems). 2 ml of cell extract were added to 50 ml of reaction buffer (Luciferase Assay Buffer from Promega Luciferase Assay System kit) in a luminometer cuvette. Light intensity was measured inside the detection linear range. Results are expressed as relative intensity units (LRU). Definitive results are expressed as relative LRU/ b-gal activities.