Promoter-dependent Effect of IKKα on NF-κB/p65 DNA Binding*

IKKα regulates many chromatin events in the nuclear phase of the NF-κB program, including phosphorylation of histone H3 and removal of co-repressors from NF-κB-dependent promoters. However, all of the nuclear functions of IKKα are not understood. In this study, using mouse embryonic fibroblasts IKKα knock-out and reexpressing IKKα after retroviral transduction, we demonstrate that IKKα contributes to NF-κB/p65 DNA binding activity on an exogenous κB element and on some, but not all, endogenous NF-κB-target promoters. Indeed, p65 chromatin immunoprecipitation assays revealed that IKKα is crucial for p65 binding on κB sites of icam-1 and mcp-1 promoters but not on iκbα promoter. The mutation of IKKα putative nuclear localization sequence, which prevents its nuclear translocation, or of crucial serines in the IKKα activation loop completely inhibits p65 binding on icam-1 and mcp-1 promoters and rather enhances p65 binding on the iκbα promoter. Further molecular studies demonstrated that the removal of chromatin-bound HDAC3, a histone deacetylase inhibiting p65 DNA binding, is differentially regulated by IKKα in a promoter-specific manner. Indeed, whereas the absence of IKKα induces HDAC3 recruitment and repression on the icam-1 promoter, it has an opposite effect on the iκbα promoter, where a better p65 binding occurs. We conclude that nuclear IKKα is required for p65 DNA binding in a gene-specific manner.

IKK␣ regulates many chromatin events in the nuclear phase of the NF-B program, including phosphorylation of histone H3 and removal of co-repressors from NF-B-dependent promoters. However, all of the nuclear functions of IKK␣ are not understood. In this study, using mouse embryonic fibroblasts IKK␣ knock-out and reexpressing IKK␣ after retroviral transduction, we demonstrate that IKK␣ contributes to NF-B/p65 DNA binding activity on an exogenous B element and on some, but not all, endogenous NF-B-target promoters. Indeed, p65 chromatin immunoprecipitation assays revealed that IKK␣ is crucial for p65 binding on B sites of icam-1 and mcp-1 promoters but not on ib␣ promoter. The mutation of IKK␣ putative nuclear localization sequence, which prevents its nuclear translocation, or of crucial serines in the IKK␣ activation loop completely inhibits p65 binding on icam-1 and mcp-1 promoters and rather enhances p65 binding on the ib␣ promoter. Further molecular studies demonstrated that the removal of chromatin-bound HDAC3, a histone deacetylase inhibiting p65 DNA binding, is differentially regulated by IKK␣ in a promoter-specific manner. Indeed, whereas the absence of IKK␣ induces HDAC3 recruitment and repression on the icam-1 promoter, it has an opposite effect on the ib␣ promoter, where a better p65 binding occurs. We conclude that nuclear IKK␣ is required for p65 DNA binding in a gene-specific manner.
Nuclear factor-B (NF-B) 5 is a key transcription factor involved in the expression of genes regulating innate and adapt-ive immunity (1), cellular proliferation, survival (2,3), and development (4). NF-B consists of homo-or heterodimers of a group of five proteins, namely NF-B1 (p50 and its precursor p105), NF-B2 (p52 and its precursor p100), p65/RelA, c-Rel, and RelB (5). In unstimulated cells, NF-B is sequestered in the cytoplasm through its tight association with inhibitory proteins of the IB family, comprising notably IB␣ (5). Upon cellular stimulation with proinflammatory cytokines, lipopolysaccharide, antigens or viral products, the so-called classical NF-B activation pathway is engaged. IB␣ is rapidly phosphorylated on Ser 32 and Ser 36 , which triggers its polyubiquitination and subsequent degradation by the 26 S proteasome. The freed NF-B then translocates into the nucleus, where it binds B sites and enhances transcription of a large number of genes implicated notably in the inflammatory response (5). Phosphorylation of IB␣ on Ser 32 and Ser 36 is achieved by the IB kinase (IKK) complex, which includes the scaffold protein NF-B essential modulator (NEMO; also called IKK␥) (6) and the IKK␣ and IKK␤ kinases (7). A novel NEMO-independent NF-B-activating pathway was recently described. This alternative pathway is engaged notably upon lymphotoxin-␤ or B cell-activating factor induction and enhances NF-B-inducing kinase-and IKK␣-dependent processing of p100 into p52 (8,9). This subunit binds DNA in association with its partners and stimulates transcription of genes important for secondary lymphoid organ development, B cell homeostasis, and adaptive immunity (1,10). IKK-independent phosphorylation of IB␣ on Tyr 42 has also been reported upon sodium pervanadate or hypoxia/reoxygenation treatment (11)(12)(13). This tyrosine phosphorylation is mediated by c-Src tyrosine kinase and triggers dissociation/degradation of IB␣ from NF-B complexes, allowing NF-B to translocate into the nucleus (12,14). Generation of knock-out mice for IKK complex subunits has revealed that IKK␤ is the main kinase responsible for IB␣ Ser 32 and Ser 36 phosphorylation and that NEMO/IKK␥ assembles the IKKs into a functional kinase complex upon the classical pathway (15)(16)(17)(18). On the other hand, whereas IKK␣ is crucial for p100 phosphorylation and subsequent NF-B activation through the alternative pathway (10), in most cases this kinase is not required for the signal-induced phosphorylation of IB␣ in the cytoplasm. Nevertheless, an optimal induction of NF-Bdependent genes through the classical pathway appears to rely on the nuclear translocation of IKK␣ (19). Indeed, once in the nucleus, IKK␣ acts in a process called derepression that allows full NF-B-mediated transcription by removing repressor complexes, such as SMRT and HDAC3, from target promoters (20). IKK␣ also phosphorylates histone H3, a component of nucleosomes. This phosphorylation triggers subsequent acetylation of histone H3 on Lys 14 by the IKK␣associated histone acetyltransferase CREB-binding protein, a crucial step in modulating chromatin accessibility at NF-B-responsive promoters (21)(22)(23). Besides a role of nuclear IKK␣ in positively regulating NF-B-dependent gene transcription, IKK␣ also limits NF-B activation in LPS-stimulated macrophages by mediating p65 and c-Rel turnover through phosphorylation of their C-terminal part (24). Some authors have also suggested that IKK␣ contributes to direct p65 DNA binding, since electrophoretic mobility shift assay (EMSA) experiments using ikk␣ Ϫ/Ϫ MEFs or HeLa cells transfected with IKK␣ small interfering RNA exhibit a clear inhibition of NF-B binding activity upon TNF-␣ treatment, despite a quite unaltered IB␣ phosphorylation/degradation (25,26). However, these results seemed inconsistent with other reports (27,28), including recent data obtained by p65 chromatin immunoprecipitation (ChIP) analysis on several NF-B target genes in ikk␣ Ϫ/Ϫ MEFs (21,23). Such a discrepancy reflects the need to determine whether or not a general mechanism can be drawn for all genes or whether some promoter-specific effects occur.
Here, we have addressed the role of IKK␣ in the control of NF-B/p65 DNA binding in mouse embryonic fibroblasts lacking IKK␣ or complemented with either wild-type (WT) or mutant IKK␣ upon TNF-␣ stimulation. We showed that IKK␣ is required for p65 DNA binding on specific, but not all NF-B-dependent, promoters. Indeed, p65 ChIP assays revealed that IKK␣ is crucial for p65 binding on B sites of the icam-1 (intercellular adhesion molecule-1) and mcp-1 (monocyte chemoattractant protein-1) promoters but not on the ib␣ promoter. This binding requires IKK␣ catalytic activity and an intact nuclear localization sequence (NLS), suggesting that IKK␣ acts in the nucleus to mediate its effects. Further molecular studies demonstrated that IKK␣ modifies HDAC3 recruitment in a promoter-specific manner, thereby explaining its gene-specific activity.

EXPERIMENTAL PROCEDURES
Cell Culture, Antibodies, and Reagents-MEFs were cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% of heat-inactivated fetal calf serum and 2 mM L-glutamine. Antibodies against p65 and HDAC3 were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) (sc-109 for p65 ChIP and Western blotting, sc-372G for p65 immunocytochemistry, and sc-11417 for HDAC3 ChIP). The antibodies against IKK␣ were from Pharmingen (for Western blotting) and Santa Cruz Biotechnology (sc-7182 for immunocytochemistry). Antibody against IKK␤ was from Upstate Biotechnology (catalog number 05-035). Other antibodies were from StressGen (HSP60) and BD Transduction Laboratories (NBS1). Antibody against IB␣ was a gift from R. Hay (St. Andrews, UK). Alexa Fluor 488 donkey antigoat and 546 goat anti-rabbit antibodies were from Molecular Probes. Human recombinant TNF-␣ was from Peprotech, trichostatin A (TSA) was from Sigma, and SYBR green PCR master mix was from Applied Biosystems. Sodium pervanadate was prepared as described (29).
Plasmids, Site-directed Mutagenesis, and Retroviral Gene Transfer-pMX retroviral vector containing human wildtype ikk␣ and ikk␣ AA were previously described (8). The IKK␣ NLS mutant was generated by site-directed mutagenesis (Stratagene) by replacing Lys 235 , Lys 236 , and Lys 237 of WT IKK␣ with alanines. These constructs were transfected by the calcium phosphate method into 293T cells in combination with the pEC ampho expression vector (8). Two days later, supernatants were collected and filtered prior to the transduction of ikk␣ Ϫ/Ϫ MEFs, and IKK␣ expression was confirmed by Western blotting.
Western Blotting and EMSA-Cytoplasmic and nuclear extracts were prepared as previously described (30). Cytoplasmic extracts were analyzed by Western blotting as described (31). Nuclear extracts and EMSA experiments were carried out as described (31) using 32 P-labeled oligonucleotide probes (Eurogentec) corresponding to the B site of the HIV-1 long terminal repeat.
Immunocytochemistry and Confocal Microscopy-p65 and IKK␣ nuclear translocation was visualized by confocal microscopy. To visualize p65, TNF-␣-treated MEFs were fixed with 4% paraformaldehyde and permeabilized with 0.3% Triton X-100. After washing with phosphate-buffered saline containing 10% fetal calf serum, slides were incubated with goat anti-p65 antibody (sc-372G; 1:100 dilution; Santa Cruz Biotechnology) for 1 h at 37°C and then washed and incubated with Alexa Fluor 488 donkey anti-goat antibody (1:400 dilution) for 1 h at 37°C. The slides were then washed, incubated with propidium iodide to visualize nuclei, mounted, and analyzed with a Leica TCS SP2 confocal microscope (Van Hopplynus). For IKK␣ immunolocalization, slides were incubated in 100% methanol at Ϫ20°C for 6 min and then washed with phosphate-buffered saline and blocked in phosphate-buffered saline containing 0.5% gelatin (Bio-Rad) and 0.25% bovine serum albumin (Invitrogen). Then cells were incubated with the primary antibody (sc-7182; 1:100 dilution; Santa Cruz Biotechnology) in blocking solution overnight at 4°C. After washing, cells were incubated with Alexa Fluor 546 goat anti-rabbit antibody (1:100 dilution) for 2 h at room temperature. Samples were mounted and analyzed with a LSM 510 Zeiss confocal microscope.
Chromatin Immunoprecipitation Assay-ChIP assays were carried out, and the solutions were prepared in our laboratory following the Upstate Cell Signaling protocol. After cross-linking with formaldehyde, treated cells were lysed and sonicated such that DNA fragments were 200 -1,000 base pairs in length. After a preclear with protein A-agarose beads saturated with herring sperm DNA (Sigma), extracts were incubated overnight with a rabbit polyclonal anti-p65 antibody (sc-109; 2 g; Santa Cruz Biotechnology) or HDAC3 antibody (sc-11417; 2 g; Santa Cruz Biotechnology). To take into account aspecific binding to the beads, a treated extract was also incubated with an irrelevant antibody (anti-FLAG; M2; Sigma). The next day, precipitation was carried out with saturated protein A-agarose beads. Cross-link was reversed at 65°C for 4 h, and precipitated DNA was purified using phenol/chloroform extraction. Quantitative PCR (using SYBR green PCR master mix; Applied Biosystems) was performed on the immunoprecipitated DNA by normalizing to input DNA for each sample. The following primers, amplifying specific B sites of the following genes, were used: icam-1 forward, 5Ј-CATTA-CTTCAGTTTGGAAATTCCTAGATC3Ј; icam-1 reverse, 5Ј-GGAACGAGGGCTTCGGTATT-3Ј; mcp-1 forward, 5Ј-CACCCCATTACATCTCTTCCCC-3Ј; mcp-1 reverse, 5Ј-TGTTTCCCTCTCACTTCACTCTGTC-3Ј; IB␣ forward, 5Ј-TGGCGAGGTCTGACTGTTGTGG-3Ј; IB␣ reverse, 5Ј-GCTCATCAAAAAGTTCC-CTGTGC-3Ј.

IKK␣ Is Required for NF-B DNA Binding on an Exogenous B Element upon Pervanadate and TNF-␣
Stimulation-To assess potential additional roles of the IKK complex subunits beside the signal-induced degradation of the IB␣ proteins in the cytoplasm, we treated MEF cells with sodium pervanadate (Pv). This treatment induces tyrosine phosphorylation and subsequent degradation of IB␣ through a IKK-independent mechanism (12,13,29). As revealed by an electromobility shift assay using a probe corresponding to the B site of the HIV-1 long terminal repeat, Pv induces a strong NF-B activation in WT MEFs, due to an almost complete degradation of IB␣ (Fig. 1A). Interestingly, Pvinduced NF-B DNA binding activity is totally abolished in MEFs lacking both IKK␣ and IKK␤, despite unaltered IB␣ degradation (Fig.  1A). The analysis of single knockout MEFs for either IKK␣ or IKK␤ revealed that the abolition of NF-B binding activity is still observable in ikk␣ Ϫ/Ϫ MEFs, but there is a partial recovery of NF-B DNA binding in ikk␤ Ϫ/Ϫ MEFs, suggesting that IKK␣ is required for NF-B DNA binding on an exogenous B element (Fig. 1A). To further confirm the role of IKK␣ in NF-B binding, we reconstituted ikk␣ Ϫ/Ϫ MEFs with retroviral expression vectors encoding either wild-type IKK␣ or the inactivable IKK␣ mutant harboring the mutations S176A and S180A within the activation loop (IKK␣ AA) (32). As control, we used ikk␣ Ϫ/Ϫ MEFs reconstituted with an empty vector. Western blot analysis confirmed the expression of IKK␣ in reconstituted MEFs (Fig. 1B). We observed a recovery in NF-B DNA binding in ikk␣ Ϫ/Ϫ MEFs reconstituted with both IKK␣ WT and AA upon Pv stimulation but not in MEFs transduced with an empty vector (Fig. 1C). The kinetics of IB␣ degradation were mostly the same in the three cell lines (Fig. 1C). These data suggest that IKK␣ contributes to the NF-B DNA binding on an exogenous B element in Pvstimulated MEFs independently of its kinase activity. The same experiments were carried out with TNF-␣ as an inducer. Mobility shift assays revealed a clear inhibition of NF-B DNA binding in ikk␣ Ϫ/Ϫ MEFs transduced with an empty vector, com- for the indicated times. Nuclear extracts were analyzed by EMSA using a radioactive probe corresponding to the B consensus sequence of the HIV-1 long terminal repeat, and IB␣ degradation was detected by an anti-IB␣ blot performed on cytoplasmic extracts. B, ikk␣ Ϫ/Ϫ MEFs were infected with retrovirus containing either empty pMX vector or pMX encoding IKK␣ WT or AA. IKK␣ expression was checked by Western blotting. IKK␤ Western blotting was also carried out as loading control. C, ikk␣ Ϫ/Ϫ MEFs infected with retrovirus containing either empty pMX vector or pMX encoding IKK␣ WT or AA were treated with Pv. NF-B activation was analyzed by EMSA with nuclear extracts, and IB␣ degradation was detected by Western blotting with cytoplasmic extracts. pared with WT MEFs, despite a similar profile of IB␣ degradation ( Fig. 2A). Reconstitution of ikk␣ Ϫ/Ϫ MEFs with IKK␣ WT or AA restored a normal NF-B binding upon TNF-␣ induction (Fig. 2B). Altogether, these results suggest that IKK␣ is required for NF-B DNA binding independently of its kinase activity.
Mutation of IKK␣ Putative Nuclear Localization Signal Abrogates NF-B DNA Binding in Vitro-Recently, a putative NLS was identified within the kinase domain of IKK␣ (33). To further explore whether IKK␣ must enter into the nucleus to modulate the NF-B DNA binding activity, we generated a mutant form of IKK␣ bearing alanines instead of three important lysines within its NLS (33). This mutant failed to accumulate in the nucleus when overexpressed in 293T cells, whereas the wild-type IKK␣ did (Fig. 3A). Next we reconstituted ikk␣ Ϫ/Ϫ MEFs with retrovirus encoding IKK␣ NLS mutant. Western blot analysis confirmed the expression of IKK␣ NLS-mutated in reconstituted MEFs (Fig. 3B). To further characterize the functionality of IKK␣ NLS in MEFs, we carried out confocal microscopy. Detection of endogenous IKK␣ in WT MEFs revealed that the cellular distribution of IKK␣ is mainly cytoplasmic, but this protein also appears in the nucleus as speckles (Fig. 3C). Similar results were obtained using ikk␣ Ϫ/Ϫ MEFs transduced with WT IKK␣ (Fig. 3C). On the contrary, no nuclear distribution was observed using the NLS-mutated IKK␣ (Fig. 3C). Absence of immunoreactivity in ikk␣ Ϫ/Ϫ MEFs confirmed antibody specificity (Fig. 3C). This suggests that a small portion of IKK␣ enters within the nucleus in MEFs, and the mutation of IKK␣ NLS prevents this translocation. We next carried out a mobility shift experiment with MEFs expressing NLS-mutated IKK␣ compared with MEFs expressing WT IKK␣. TNF-␣ and Pv-induced NF-B DNA binding was strongly reduced in MEFs expressing NLS-mutated IKK␣, despite a quite unaltered IB␣ degradation (Fig. 3D). These data suggest that a small fraction of IKK␣ functions in the nucleus enhancing NF-B DNA binding.
p65 Translocates Normally into the Nucleus in ikk␣ Ϫ/Ϫ MEFs or MEFs Expressing NLS-mutated IKK␣ upon TNF-␣ Stimulation-To explore whether the absence of NF-B DNA binding in MEFs lacking IKK␣ or expressing IKK␣ NLS-mutated is due to an altered NF-B nuclear translocation, we performed a large panel of anti-p65 Western blots on cytoplasmic and nuclear extracts of MEFs treated with TNF-␣. As shown in Fig. 4A, p65 nuclear accumulation is maximal after 30 min of treatment in WT MEFs and then decreases at 60 min. The same profile of p65 nuclear translocation was observed in ikk␣ Ϫ/Ϫ MEFs reconstituted with an empty vector or expressing IKK␣ WT or AA (Fig. 4A). p65 nuclear translocation was also observed in MEFs where the IKK␣ NLS was mutated. Interestingly, the basal level of nuclear p65 is more important compared with other cells (Fig. 4A). Western blot analysis for cytoplasmic p65 revealed that its level of expression is constant across the analyzed cell lines (Fig. 4A). The purity of extracts was controlled by reprobing the membrane with antibodies raised against NBS and HSP60 (Fig. 4A). Unaltered p65 nuclear accumulation upon TNF-␣ stimulation of MEFs lacking IKK␣ or expressing IKK␣ NLS-mutated was further confirmed by immunolocalization of p65 using confocal microscopy (Fig. 4B). This experiment also confirmed an elevated basal nuclear accumulation of p65 in MEFs IKK␣ NLS-mutated (Fig. 4B). Altogether, we conclude that inhibition of NF-B activation observed by EMSA in the absence of IKK␣ or when IKK␣ NLS is mutated is not due to a lack of p65 translocation.
IKK␣ Mediates Recruitment of p65 on Specific NF-Bdependent Endogenous Promoters-To extend our analyses, we used p65 ChIP assays to determine whether defects in NF-B DNA binding observed by electromobility shift assays were also confirmed on endogenous promoters. After chromatin immunoprecipitation with p65 antibody, real time PCR was used to amplify B consensus sequences of promoters of three NF-B target genes: icam-1, mcp-1, and ib␣. These genes were selected because they encode proteins important in inflammation and innate immunity (icam-1 and mcp-1) or for the negative feedback of NF-B regulation (ib␣). TNF-␣ treatment of WT MEFs induced two waves of p65 recruitment on icam-1 and mcp-1 promoter B sites, which reached a maximum at 30 and 120 min. However, p65 recruitment is clearly inhibited in ikk␣ Ϫ/Ϫ MEFs, particularly after 30 and 60 min of treatment (Fig. 5, A and B). Interestingly, p65 recruitment on ib␣ promoter in ikk␣ Ϫ/Ϫ MEFs is slightly reduced but not totally inhibited, as observed for icam-1 and mcp-1 (Fig. 5C). Similar results were obtained with the interleukin-6 promoter (data not shown). These data suggest that IKK␣ is required for p65 DNA binding on some but not all promoters. Reconstitution of ikk␣ Ϫ/Ϫ MEFs with a vector  A, pMX retroviral vector encoding IKK␣ wild-type or NLS-mutated was transfected into 293T cells. Nuclear and cytoplasmic extracts were then prepared, and cellular distribution of IKK␣ was analyzed by Western blotting. The purity of extracts was controlled by probing the membrane with antibodies raised against NBS and HSP60, a nuclear and mitochondrial protein, respectively. B, ikk␣ Ϫ/Ϫ MEFs were infected with retrovirus containing pMX vector encoding IKK␣ WT or NLS. IKK␣ expression was checked by Western blotting. IKK␤ Western blotting was also carried out as loading control. C, IKK␣ cellular distribution was visualized by confocal microscopy in MEFs that were WT, ikk␣ Ϫ/Ϫ , or ikk␣ Ϫ/Ϫ transduced with a vector coding for IKK␣ WT or NLS-mutated. Diagrams depict the intensity of the fluorescence for each cell type along lines drawn across the nucleus of the cells. NE, nuclear envelope. D, ikk␣ Ϫ/Ϫ MEFs transduced with a vector coding for IKK␣ WT or NLS-mutated were treated with TNF-␣ (200 units/ml) or Pv (200 M) for the indicated times. NF-B activation and IB␣ degradation were studied as in Fig. 1.

FIGURE 4. p65 translocates normally into the nucleus upon TNF-␣ stimulation in MEFs lacking IKK␣ or expressing IKK␣ NLS-mutated.
A, MEFs that were wild-type or transduced with either an empty vector or vector coding for IKK␣ WT, AA, or NLS-mutated were treated with TNF-␣ (200 units/ml) for the indicated times. Nuclear (N) and cytoplasmic (C) extracts were prepared and analyzed by Western blotting using p65 and IKK␣ antibodies. The membrane was then stripped and reprobed with antibodies against HSP60 and NBS1 to check the purity of the extracts. B, MEFs that were wild-type or transduced with either an empty vector or vector coding for IKK␣ WT, AA, or NLS-mutated were treated with TNF-␣ for the indicated times. p65 immunofluorescence was then carried out as described under "Experimental Procedures" with a specific antibody (green). Nuclei were visualized using propidium iodide staining (red). Ctrl, control.
encoding WT IKK␣ completely restored recruitment of p65 on icam-1 and mcp-1 promoters, whereas no changes were observed regarding p65 DNA binding on ib␣ promoter (Fig. 6, A-C). Complementation of ikk␣ Ϫ/Ϫ MEFs with IKK␣ AA or NLS clearly inhibits p65 recruitment on icam-1 and mcp-1 promoters but not on the ib␣ promoter (Fig. 6, A-C). In this case, an increase in p65 DNA binding is even observed (Fig. 6C). Collectively, these experiments prompt us to conclude that IKK␣ is crucial for p65 DNA binding on some but not all promoters. Indeed, the ib␣ promoter does not seem to require IKK␣ for p65 recruitment, whereas icam-1 and mcp-1 promoters do. p65 recruitment on these promoters is abrogated by IKK␣ AA or NLS mutations, whereas these mutations rather enhance p65 binding on the ib␣ promoter.

Inhibition of p65 Binding on icam-1 and mcp-1 Promoters in ikk␣ Ϫ/Ϫ MEFs Is Relieved by TSA Pretreatment-Acetylation
has been described as a critical posttranslational mechanism regulating p65 DNA binding activity (34,35), and treatment of cells with TSA, an HDAC inhibitor, potentiated NF-B activation induced by TNF-␣ (36). This observation suggests that inhibition of HDAC activity is critical for a correct NF-B DNA binding. Therefore, we wanted to explore whether TSA pretreatment modifies p65 DNA binding in ikk␣ Ϫ/Ϫ MEFs treated with TNF-␣. TSA restored normal p65 recruitment on icam-1 and mcp-1 promoters in ikk␣ Ϫ/Ϫ MEFs (Fig. 7, A and B), reaching values similar to those obtained in WT MEFs (see Fig. 5). Conversely, TSA pretreatment did not increase p65 binding on the ib␣ promoter (Fig. 7C). Altogether, these data suggest that TSA can substitute for IKK␣ for NF-B DNA binding on specific promoters.

HDAC3 Recruitment on icam-1 and ib␣ Promoters Is Oppositely
Regulated by IKK␣-The results obtained with TSA pretreatment led us to further explore the role of HDACs in IKK␣-mediated p65 DNA binding. HDAC3, a class I histone deacetylase, has been reported to directly deacetylate p65, thereby inhibiting its DNA binding (34). HDAC3 is associated with NF-B-dependent promoters and negatively regulates NF-Bdependent transcription together with co-repressors, such as SMRT (20). Using laminin attachment to activate NF-B, it was reported that IKK␣ directly removes SMRT and HDAC3 from promoters, thereby allowing NF-B-mediated transcription (20). Thus, we studied the recruitment of HDAC3 on the promoters of two genes differentially regulated by IKK␣ (i.e. icam-1 and ib␣) upon TNF-␣ stimulation. In ikk␣ Ϫ/Ϫ MEFs expressing WT IKK␣, TNF-␣ induces a slight removal of HDAC3 from the icam-1 promoter (Fig. 8A). On the contrary, HDAC3 is dramatically recruited in MEFS lacking IKK␣ and reconstituted with IKK␣ AA or NLS-mutated (Fig. 8A). No change in chromatin-bound HDAC3 was observed on the ib␣ promoter in WT MEFs stimulated with TNF-␣ (Fig. 8B), whereas a dramatic removal is induced in ikk␣ Ϫ/Ϫ MEFs or MEFs expressing IKK␣ AA or NLS-mutated (Fig. 8B). This FIGURE 5. IKK␣ mediates recruitment of p65 on specific NF-B-dependent endogenous promoters. MEFs that were wild-type or IKK␣-deficient were treated with TNF-␣ for the indicated times. CHiP assays using a p65 antibody were performed as described under "Experimental Procedures." Real time PCR was then carried out on immunoprecipitated DNA using primers amplifying B sites of promoters of icam-1 (A), mcp-1 (B), and ib␣ (C) genes.
HDAC3 removal is not induced by the prior p65 DNA binding, since it is still observable in p65 KO MEFs stimulated with TNF-␣ (data not shown). Collectively, these results suggest that IKK␣ oppositely regulates the dynamic of HDAC3 recruitment on icam-1 and ib␣ promoters.

DISCUSSION
The study of nuclear events involved in NF-B activation is the focus of intensive research in many laboratories. IKK␣, one of the subunits of the IKK complex, has recently been associated with NF-B nuclear action, notably through its phosphorylation of histone H3, a component of nucleosomes (21,23). However, the precise nuclear functions of IKK␣ are poorly understood. Here, the use of IKKindependent pathways triggering IB␣ degradation (i.e. sodium pervanadate stimulation) allowed us to study the precise nuclear roles of IKK subunits in MEFs defective for IKK␣ and IKK␤. We found that IKK␣ is crucial for NF-B DNA binding on an exogenous B element, whereas IKK␤ seems less important. TNF-␣ stimulation gave rise to the same results. This is in agreement with other studies demonstrating that the absence of IKK␣ inhibits NF-B DNA binding activity in HeLa cells or mouse embryonic fibroblasts (25,26). The complementation of ikk␣ Ϫ/Ϫ MEFs with either WT or inactivable (AA) IKK␣ totally rescued NF-B DNA binding upon Pv and TNF-␣ stimulation, suggesting that IKK␣ kinase activity is not required for this function. Meanwhile, mutation of important residues in a putative NLS within IKK␣, thereby preventing its nuclear translocation, dramatically inhibited NF-B DNA binding on an exogenous B element. This suggests that IKK␣ acts in the nucleus to enhance NF-B DNA binding. Indeed, we clearly observed a nuclear distribution of IKK␣ as speckles in WT MEFs or ikk␣ Ϫ/Ϫ MEFs reexpressing IKK␣ WT, whereas mutation of IKK␣ NLS prevented this accumulation. We thus propose that a small portion of IKK␣ enters the nucleus to regulate chromatin events. This result seems to conflict with another work reporting an important nuclear accumulation of IKK␣ after TNF-␣ stimulation (21). We never observed such a massive translocation upon TNF-␣ treatment (data not shown). These discrepant results probably reflect technical differences between laboratories in the design of experiments. For example, the use of ikk␣ Ϫ/Ϫ cells to take into account aspecific binding of antibodies appears highly desirable to avoid artifactual results. In a more general context, IKK␣ nuclear translocation is still a matter of debate in literature. Whereas some researchers observe high levels of IKK␣ in the nucleus of various cell types (21,23,33,37), others fail to detect any nuclear IKK␣ and even use this protein as a negative control in their ChIP experiments (38). These discrepancies probably reflect cell type-specific functions of IKK␣.
Confocal microscopy experiments and p65 blots on cytoplasmic and nuclear fractions clearly demonstrated that the absence of NF-B DNA binding in MEFs defective for IKK␣ or expressing IKK␣ NLS-mutated was not due to an inhibition of p65 nuclear translocation. Interestingly, MEFs expressing IKK␣ NLS-mutated exhibit a constitutive nuclear accumulation of p65, suggesting that IKK␣ regulates p65 nucleo-cytoplasmic shuttling. Recently, Lawrence et al. (24) highlighted a role for IKK␣ in accelerating promoter clearance of p65 in macrophages, thereby contributing to the resolution of inflammation. This could explain the enhanced p65 nuclear accumulation observed in MEFs IKK␣ NLS-mutated, although, in this case, p65 is not found to be associated with target promoters.
In a second part, we extended our EMSA analyses to p65 ChIP assays. These experiments confirmed the dramatic decrease of p65 binding on icam-1 and mcp-1 promoters, two NF-B target genes, in IKK␣ KO MEFs stimulated with TNF-␣. On the contrary, the ib␣ promoter does not seem to require IKK␣ for p65 binding, highlighting a promoter-specific function for IKK␣. Normal p65 binding on the ib␣ promoter in the absence of IKK␣ was also reported by other groups (21,23), suggesting that this promoter behaves differently from icam-1 and mcp-1. p65 ChIP also confirmed the crucial role of IKK␣ NLS sequence for p65 binding, suggesting that IKK␣ must enter into the nucleus to achieve its roles. Interestingly, IKK␣ catalytic activity seems to be also required for p65 DNA binding on endogenous icam-1 and mcp-1 promoters, whereas EMSA experiments revealed no change in NF-B binding on an exogenous B element using MEFs expressing IKK␣ AA. This suggests that IKK␣ has a dual function in p65 DNA binding. First, the requirement of nuclear IKK␣ for direct p65 binding on both exogenous B elements and endogenous promoters suggests that IKK␣ directly modifies p65 to allow its DNA binding. This modification is likely to be acetylation, since IKK␣ has been recently reported to interact with CREB-binding protein, a histone acetyltransferase capable of acetylating p65 and thus enhancing its DNA binding (23,34). IKK␣ might thus act as a nuclear scaffold protein required for the post-translational modifications of transcription factors, thereby enhancing their DNA binding capacities. The second role of IKK␣ involves its catalytic activity in the chromatin context. Hoberg et al. (20) recently reported that IKK␣ activity was necessary to phosphorylate the chromatin-bound corepressor SMRT, thereby inducing its removal together with HDAC3. This mechanism, deciphered upon NF-B activation induced by laminin attachment, is likely to be the same in our system. Indeed, we observed a dramatic recruitment of HDAC3 on the icam-1 promoter in MEFs lacking IKK␣ or expressing IKK␣ AA or NLS-mutated stimulated with TNF-␣. One hypothesis is that promoter-bound HDAC3 inhibits p65 binding on the icam-1 promoter by inducing its deacetylation, since this inhibition is relieved by TSA pretreatment. To further confirm this hypothesis, we transduced p65 KO MEFs with retroviruses encoding WT p65 or a hypoacetylated mutant containing a lysine to arginine substitution at the position 221, one major acetylation site of p65 (34). As already described, we found out that p65 K221R does not bind DNA due to a defect in nuclear translocation (data not shown). This