Inhibition of NF-κB Activation by Peptides Targeting NF-κB Essential Modulator (NEMO) Oligomerization*

NF-κB essential modulator/IKK-γ (NEMO/IKK-γ) plays a key role in the activation of the NF-κB pathway in response to proinflammatory stimuli. Previous studies suggested that the signal-dependent activation of the IKK complex involves the trimerization of NEMO. The minimal oligomerization domain of this protein consists of two coiled-coil subdomains named Coiled-coil 2 (CC2) and leucine zipper (LZ) (Agou, F., Traincard, F., Vinolo, E., Courtois, G., Yamaoka, S., Israel, A., and Veron, M. (2004) J. Biol. Chem. 279, 27861–27869). To search for drugs inhibiting NF-κB activation, we have rationally designed cell-permeable peptides corresponding to the CC2 and LZ subdomains that mimic the contact areas between NEMO subunits. The peptides were tagged with the Antennapedia/Penetratin motif and delivered to cells prior to stimulation with lipopolysaccharide. Peptide transduction was monitored by fluorescence-activated cell sorter, and their effect on lipopolysaccharide-induced NF-κB activation was quantified using an NF-κB-dependent β-galactosidase assay in stably transfected pre-B 70Z/3 lymphocytes. We show that the peptides corresponding to the LZ and CC2 subdomains inhibit NF-κB activation with an IC50 in the μm range. Control peptides, including mutated CC2 and LZ peptides and a heterologous coiled-coil peptide, had no inhibitory effect. The designed peptides are able to induce cell death in human retinoblastoma Y79 cells exhibiting constitutive NF-κB activity. Our results provide the “proof of concept” for a new and promising strategy for the inhibition of NF-κB pathway activation through targeting the oligomerization state of the NEMO protein.

ing inflammatory responses, antigenic stimulation of B and T lymphocytes, viral infection, oncogenesis, and apoptosis (1)(2)(3)(4). In resting cells, NF-B transcription factors are sequestered in the cytoplasm via their association with IB proteins. Stimulation of cells by several factors, including the cytokines tumor necrosis factor-␣ and interleukin-1, and lipopolysaccharides (LPSs), results in the activation of the IB kinase (IKK) complex. This complex is composed of two protein kinases (IKK-␣ and IKK-␤) and a non-catalytic regulatory protein called NEMO (NF-B essential modulator) (5,6). Upon activation, the IKK complex phosphorylates the IB proteins, leading to their ubiquitination and proteasome degradation. The released NF-B transcription factors are then translocated into the nucleus. NEMO plays a critical role in the activation of this pathway. Indeed, the inability of NEMO-deficient cells to activate NF-B genes in response to extracellular stimuli can be overcome by transfection with the wild-type NEMO gene (5).
The biochemical mechanisms triggering IKK activation in response to proinflammatory stimuli remain obscure. It was recently reported that NEMO is phosphorylated and ubiquitinated upon tumor necrosis factor-␣ stimulation (7)(8)(9), but it remains to be established whether these post-translational modifications contribute to the activation of the IKK complex. Oligomerization of NEMO is necessary for kinase activation; enforced oligomerization of NEMO leads to constitutive activation of the IKK complex (10 -12).
Its amino acid sequence suggests that the NEMO protein consists of several domains. The N-terminal part of the polypeptide contains a large coiled-coil motif (CC1) and all the residues involved in the interaction of the protein with the IKK kinases (IKK-binding domain, see Fig. 1) (13). The C-terminal half (residues 250 -412) is composed of two successive coiledcoil motifs, CC2 (residues 253-285) and LZ (residues 301-337), and a zinc finger motif (ZF) at the extreme C terminus of the polypeptide. It is this C-terminal part of NEMO that is involved in the oligomerization of the protein (14,15). The C-terminal region is essential for protein function as NF-B activation in NEMO-deficient cells cannot be restored by complementation with a gene carrying a mutation in the oligomerization domain (16). In addition, the mutations responsible for the NEMOassociated pathologies, incontinentia pigmenti and anhidrotic ectodermal dysplasia with immunodeficiency, are found within the gene encoding this C-terminal part of the protein (17)(18)(19).
The design of inhibitors able to block the NF-B pathway is of obvious interest, and targeting NEMO is a promising approach because this protein is a central and non-redundant component of the IKK complex. In general, peptides designed to affect the function of the protein by interfering with proteinprotein interactions have long been considered, but concerns over their low bioavailability and stability have limited their development (20). However, there is currently renewed interest in developing peptides for use as drugs (21,22), particularly when the protein targets are signaling proteins exhibiting flexible and dynamic binding properties (for a review, see Ref. 23). Therapeutic peptides have already been successfully designed, such as the gp41 peptide involved in HIV-1 penetration into cells (24,25).
We have shown that the minimal oligomerization domain required for NEMO trimerization is comprised of both the CC2 and LZ coiled-coil subdomains and proposed a model for the organization of these subdomains within the oligomer (16). In this study, we report a specific inhibition by low concentrations of peptides designed to interfere with the oligomeric structure of the protein. The potential use of these peptides as therapeutic drugs for future clinical applications is discussed.

MATERIALS AND METHODS
Cell Lines, Stable and Transient Transfections, and Conditions for FACS Analysis-Stable cell lines (70Z/3-C3) were obtained after electroporation of 70Z/3 cells (26) in the presence of a cx12lacZ-B plasmid containing three tandem copies of an NF-B-binding site in the interleukin-2 promoter (27) upstream from the lacZ reporter gene (a gift from G. R. Crabtree). The human retinoblastoma cell line Y79 (American Type Culture Collection; Manassas, VA), was grown in RPMI 1640 medium supplemented with 100 units/ml penicillin, 100 g/ml streptomycin, and 10% fetal calf serum. Jurkat T cells were transiently transfected by a DEAE-dextran method with an SRE-Luc reporter plasmid as described (26). 24 h after transfection, cells were incubated without or with NEMO-derived peptides for 2 h and then mock-stimulated or stimulated with 100 ng/ml phorbol 12-myristate 13-acetate and 1 g/ml ionomycin for 5 h. Luciferase expression was carried out as described in Ref. 26. To analyze the internalization of the peptides, 70Z/3-C3 pre-B lymphocytes (0.5 ϫ 10 6 cells in 0.5 ml of culture medium) were incubated at 37°C for different lengths of time with various peptide concentrations. The cell suspension was centrifuged (400 ϫ g for 7 min at room temperature), and the resulting cell pellet was washed three times with 1 ml of PBS buffer before being resuspended in 500 l of a PBS buffer containing 0.1% sodium azide. Fluorescence analysis of the cells was performed with a FACSCalibur flow cytometry system (BD Biosciences). A minimum number of 15,000 counts were made per sample. All experiments were performed in duplicate.
Peptide Synthesis and Purification-Peptides shown in Table I as well as the long peptide sp-CC2-LZ (residues 253-336 in the mouse sequence) were synthesized as described previously (28) by using a continuous-flow Fmoc (N-(9-fluorenyl)methoxycarbonyl)/terbutyl strategy on an Applied Biosystems (Foster City, CA) Pioneer peptide synthesizer. All chemical reagents were obtained from Applied Biosystems. Peptides were N-terminally blocked with an acetyl group and C-terminally blocked with an amide. As the peptide sequences contained no cysteine residues, a single cysteine was incorporated at the N terminus to allow N-terminal specific labeling (see Table I). Biot-LZ peptides were biotinylated using EZ-link (Pierce). In cell death experiments (Fig.  9), the peptides were pretreated with iodoacetamide to prevent oxidation of the incorporated cysteine residue. Crude peptides were purified by reverse-phase MPLC using a preparative column packed with Nucleoprep (20-m particle size, 100-Å pore size) C18 100 Å and a linear gradient of acetonitrile (1%/min) in 0.08% aqueous trifluoroacetic acid (pH 2). The run time was 60 min at a flow rate of 18 ml/min. To improve the purity of the peptides, reverse-phase HPLC was repeated using a semipreparative column packed with nucleosil C18 (5 m particle size, 300 Å pore size). The BODIPY® FL N-(2-aminoethyl) maleimide fluorophore (Molecular Probes) was conjugated to the incorporated N-terminal cysteine by incubation of equimolar concentrations of the peptide and fluorophore in 50 mM ammonium acetate buffer (pH 6) for 30 min in the dark. Free fluorophore was then removed by reversephase MPLC using a preparative column packed with Nucleoprep (20-m particle size, 100-Å pore size) C18 100 Å. Peptide integrity and coupling efficiency were analyzed by mass spectrometry. Extinction coefficients were determined at 280 and 505 nm by using peptide concentrations resulting from amino acid analysis The absorbance ratio was then monitored periodically to assess the stability of the labeling. Peptides were stored at a concentration of 2 mM in water at Ϫ20°C.
NF-B Inhibition Assays-In Protocol A, 2.2 ϫ 10 5 70Z/3-C3 cells in 220 l of RPMI 1640 medium supplemented with 10% fetal calf serum and 50 M ␤-mercaptoethanol (referred to as the complete medium) were incubated with 0 -20 M peptide. After 2 h at 37°C, 100 l of cell samples were transferred in duplicate to a 96-well microtiter plate. One well was treated for 5 h with 15 g/ml LPS from Salmonella abortus (Sigma), and the control well was left untreated. After 5 h at 37°C, cells were centrifuged at 400 ϫ g for 5 min at room temperature, the pellets were washed three times with cold PBS (250 l), and the cells were lysed in a 25 mM Tris-phosphate buffer, pH 7.8, containing 8 mM MgCl 2 , 1 mM dithiothreitol, 1% Triton X-100, 15% glycerol (LB buffer), and a protease inhibitor mixture (Roche Applied Science). The lysate was centrifuged at 1000 ϫ g for 20 min at 4°C, and the supernatant was kept on ice before performing the ␤-galactosidase assay. Assays were performed on 30 l of the supernatant and in an assay mix containing 4 l of Galacton-star chemiluminescent substrate (29) and 200 l of reaction buffer (Clontech). The activity was measured with a plate luminometer (Berthold). In Protocol B, 70Z/3-C3 cells were incubated for 2 h with the peptide. Following this incubation, cells were centrifuged at 400 ϫ g for 7 min at room temperature, and the resulting pellets were washed three times with 200 l of PBS. The cell pellet was diluted three times with complete medium, after which the cells were allowed to continue their growth for at least 24 h before being stimulated with LPS. All subsequent steps were carried out as described in Protocol A.
Analytical Gel Filtration-The oligomeric state of the peptides was determined by the gel filtration of 500 l of samples on a Superdex 75 HR 10/30 column equilibrated using a 50 mM Tris-HCl buffer, pH 8. Fluorescence Anisotropy-Anisotropy measurements were performed with a PTI Quantamaster fluorometer, equipped with polarizing filters for excitation and emission, and using a photomultiplier tube in the L-configuration. All experiments were carried out in a 1-cm path length cuvette at 22°C, with excitation and emission wavelengths at 495 and 520 nm. The band pass of the excitation and emission wavelengths was 2 and 4 nm, respectively. Steady-state fluorescence anisotropy, expressed as millianisotropy units was measured as described by (16). All measurements were carried out in a 50 mM Tris-HCl buffer, pH 8, containing 150 mM KCl. Experiments were performed at least in duplicate, and each data point is the result of 20 recordings taken over a 2-min period. We verified that the effect of the inner filter was negligible at the peptide concentration used. Prior to anisotropy measurement, BA-CC2 (1 M) was preincubated overnight at 22°C, alone or with increasing concentrations (1-125 M) of sp-CC2. BA-LZ (100 nM) was preincubated overnight at 22°C alone or with 10 or 100 M concentrations of sp-CC2 (see the legend for Fig. 6). The K D was estimated by fitting the anisotropy data to the binding isotherm equation using the KaleidaGraph non-linear regression software. The binding stoichiometry (n), which is the number of bound sp CC2 per mol of BA-CC2, was estimated from the intersection of the lines (see Fig. 6, dashed lines) fitted onto the linear and plateau regions of the graph generated from the anisotropy data (30).
Competitive Titration of the Untagged sp-CC2-LZ with sp-CC2 and sp-LZ Peptides-The recombinant His-tagged CC2-LZ corresponding to the minimal oligomerization domain of NEMO was purified as described earlier (16). The synthetic sp-CC2-LZ and recombinant (His-CC2-LZ) oligomers were incubated overnight at 4°C in a 20 mM Tris-HCl buffer at pH 8.0 containing 150 mM potassium chloride and 2 M potassium thiocyanate to destroy their trimeric structure and mixed together in a final volume of 0.24 ml at a final concentration of 1 mg/ml and 0.5 mg/ml for sp-CC2-LZ and His-tag-CC2-LZ, respectively. After an extensive dialysis at 4°C overnight against a 25 mM Tris-HCl buffer, pH 8.0, containing 10 mM imidazole, 50 mM potassium chloride, and 10% glycerol (buffer A), 20 l of the mixture was incubated at 4°C with 50 l of Ni-NTA magnetic beads (Qiagen) to immobilize the heterotrimer composed of His-tagged and untagged subunits. After separation with a magnet, beads were washed three times at 4°C with 2 volumes of buffer A without (control) or with variable concentrations of sp-CC2 (WT), sp-CC2 (Mut), sp-LZ (WT), or sp-LZ (Mut). The amount of sp-CC2-LZ released was evaluated by SDS-PAGE analysis with Coomassie Blue staining. As staining and destaining of small peptides can result in diffusion out of the gel, the competitive titration was evaluated by monitoring the displacement of sp-CC2-LZ rather than by measuring the staining intensity of bound peptides.
Identification of Proteins Bound to Biotinylated LZ Peptides-70Z/ 3-C3 cells were seeded at a density of 10 6 /ml in 2 ml in 6-well plates and incubated for 2 h at 37°C without or with 10 M biotinylated LZ peptides. Cells were then extensively washed two times with cold PBS to remove excess peptide and lysed in 100 l of LB buffer containing a protease inhibitor mixture (Roche Applied Science). Lysates were clarified at 15,000 ϫ g at 4°C for 20 min, and aliquots of 10 l (10% of input) were withdrawn before incubating 50 l of magnetic streptavidin beads (Novagen) for 30 min at 4°C with the remaining materials (180 g of protein each). Biotinylated peptides were pulled down with streptavidin beads and washed twice with LB buffer. Bound proteins and clarified lysates were then analyzed by Western blotting using anti-NEMO (5) and anti-NRP (gift of Dr R. Weil) antibodies.
Cell Death Assays-Cell death was quantified using the MTS assay from Promega (CellTiter 96 ® AQ ueous one solution cell proliferation assay). To increase the sensitivity of the assay, we used BODIPY-free peptides as the absorption spectrum of the fluorophore interferes with that of the formazan generated in the assay. The assay was carried out as follows: 50 l (0.1-20 M final concentration) of the relevant peptide was added to 450 l of a Y79 cell culture (0.3 ϫ 10 6 ). As a control, a cell sample was left untreated in serum-free RPMI medium at 37°C. After incubation for either 1 or 14 h, 200 l (0.12 ϫ 10 6 cells) of the assay culture was transferred to a 96-well plate and mixed with 40 l of MTS solution. The amount of formazan produced by viable cells was measured 2 h later at 490 nm. Cell survival, estimated by observation of the untreated cells under the microscope, was expressed as the percentage of OD 490 nm . The background level of fluorescence in the reaction was caused by the cell-free RPMI medium. All experiments were repeated twice, and samples were tested in duplicate in each experiment.

Rational Design of Peptides from the Minimal Oligomerization Domain of NEMO-
We have identified the minimal oligomerization domain required for the formation of the NEMO trimer. It consists of residues 251-337 of the C-terminal half of the sequence (16) (Fig. 1A). Our model predicts that the CC2-LZ domain forms a six-stranded helical bundle composed of a trimer of CC2 domains surrounded by three LZ coiled-coils (16). This model was used to design the sp-CC2 and sp-LZ peptides, which contain residues 253-287 and residues 294 -336, respectively. These peptides were designed to interfere with the oligomerization of NEMO (Table I) tension derived from the Antennapedia/Penetratin protein, which acts as a peptide internalization vector, was added to the N terminus of each peptide. The fluorescent chromophore BODIPY was conjugated to an incorporated N-terminal cysteine to monitor peptide uptake (Table I). For clarity, we refer to the peptides labeled with BODIPY-Antennapedia as BA-CC2 and BA-LZ and to the unlabeled peptides as sp-CC2 and sp-LZ.
Cellular Uptake of NEMO-derived Peptides-We analyzed the uptake of BA-CC2, BA-LZ, and two controls (free BODIPY and BODIPY-conjugated BSA) into 70Z/3-C3 cells, which had been preincubated with the samples for 2 h, by FACS ( Fig. 2A shows the cellular uptake of BA-CC2). The fluorescence patterns of the four peptides were identical (Fig. 2B, BA-pp peak), indicating that their intracellular concentrations were similar. Free BODIPY or BODIPY-conjugated BSA was not internalized (Fig. 2B). The autofluorescence of untreated cells was similar to the fluorescence of cells treated with BODIPY-BSA or with free BODIPY (Fig. 2B), indicating that the washing protocol eliminated any dye adhering to the cell surface. The kinetics and concentration dependence of BA-CC2 uptake were linear (not shown). The intracellular concentration of BA-CC2 reached a maximum after 30 min in the presence of 20 M peptide and was unchanged after a further incubation of 5 h, the time required to monitor LPS-induced NF-B activation by our ␤-galactosidase reporter assay.
Inhibition of LPS-induced NF-B Activation by BA-CC2 and BA-LZ-To analyze the effect of BA-CC2 and BA-LZ on LPSinduced NF-B activation, murine pre-B 70Z/3 cells were stably transfected with the p12XlacZ-B plasmid containing the ␤-galactosidase reporter gene under the control of the NF-B promoter (see "Material and Methods"). A 100-fold increase in the ␤-galactosidase activity was observed after a 5-h incubation of the transfected cells with 15 g/ml LPS. Incubation of cells with any of the NEMO-derived peptides strongly reduced this LPS-induced ␤-galactosidase activity (Fig. 3). The inhibition was around 2-fold with BA-CC2 and reached around 50-fold with BA-LZ, relative to cells not treated with any peptide. The  Table I   a An N-terminal extra cysteine residue was added to each Antennapedia sequence (A) for coupling with BODIPY (B). The Antennapedia sequence is in bold, and residues predicted to belong to coiled-coil motifs are underlined. In the mutant peptides (Mut), mutations are indicated in small letters.
IC 50 values for BA-CC2 and BA-LZ were 22 and 3 M, respectively. No alterations in ␤-galactosidase activity were found in cells incubated with either BODIPY-labeled or unlabeled Penetratin (not shown). The untagged sp-LZ and sp-CC2 peptides had no effect on cell activation. We verified that the peptides were not cytotoxic using the MTS assay (data not shown). Importantly, the basal ␤-galactosidase activity measured in the absence of LPS was not significantly affected by either of the two peptides (Fig. 3, inset).

Inhibition of NF-B Activation Is Mediated by Specific Interaction of the Peptides with CC2 and LZ Coiled-Coils-BA-CC2
and BA-LZ may inhibit NF-B activation by disrupting the association between CC2 and LZ coiled-coils. We investigated the mechanism of BA-CC2 and BA-LZ inhibition using mutant  (Table I). A coiled-coil is classically represented as a "helical wheel" with the interface involving positions a and d of the heptad repeat mostly made of hydrophobic amino acids (31) (Fig. 4). Thus, substitution of a residue in the a position with a proline or glycine in CC2 should have a destabilizing effect in the ␣-helical structure. The substitution of the leucine at position d in LZ with a serine should destabilize the hydrophobic core without disrupting the ␣-helical structure. We synthesized a variant of BA-CC2 (BA-CC2 Mut) containing two Leu 3 Gly substitutions and one Ile 3 Gly substitution at the a position (Fig. 4A, top helical wheel) and a BA-LZ variant peptide (BA-LZ Mut) containing two Leu 3 Ser mutations at the d position (Fig. 4A, middle helical wheel) (Table I). BA-CC2 Mut had no effect on cell activation, and BA-LZ Mut inhibited the response to LPS by only 15%, as compared with 80% for the wild-type peptide (Fig. 4C).
A coiled-coil peptide mimicking the leucine zipper of GCN4 (BA-GCN4) was used as an additional control (Table I). Although the sequence similarity of this peptide to the LZ domain of NEMO is only 22% (Fig. 4B), the helical wheel representation shows that the leucine residues, which are major contributors to the stability of coiled-coil oligomers, are present at position d in both peptides (Fig. 4A, bottom helical wheel). In contrast, position a, known to be important for the specificity of coiled-coil interactions (31), is occupied by hydrophobic residues and a non-typical Asn in GCN4, whereas two charged amino acids (Arg and Lys) are present at this position in LZ. The BA-GCN4 peptide did not inhibit NF-B activation (Fig.  4C), despite the fact that it was internalized in manner similar to the other peptides used in this study (not shown).
The Antennapedia Sequence Results in the Dissociation of CC2 and LZ Peptide Oligomers-To increase our understanding of the mechanism by which BA-CC2 and BA-LZ inhibit NF-B activation, we analyzed their oligomeric state by gel filtration. Antennapedia-tagged peptides were compared with BODIPY-free untagged peptides (Fig. 5). At 10 M, the unlabeled sp-CC2 and sp-LZ peptides oligomerize into a trimer and a dimer, respectively (Fig. 5, lower panel) (16). In contrast, both tagged peptides elute at a volume corresponding to their monomeric form, showing that the addition of the N-terminal Antennapedia sequence alters the homotypic coiled-coil interactions of BA-CC2 and BA-LZ. The mutant peptide sp-CC2 Mut also eluted at a volume corresponding to its monomeric form. Dimers were formed by the sp-LZ Mut peptide; however, dimer formation was markedly reduced as compared with that of the native peptide (Fig. 5, bottom panel, dashed line).
Interactions of BA-CC2 and BA-LZ with sp-CC2-The BA-CC2 and BA-LZ peptides were unable to self-associate. We studied whether they could form a heterologous complex with the trimeric CC2 protein by fluorescence polarization. Fig. 6 shows the binding isotherm of sp-CC2 to BA-CC2, under conditions where the amount of BA-CC2 was constant. A K A of 15.2 M was calculated using a stoichiometry of 0.8, estimated from the intercept of the tangent of the initial part of the curve with the asymptote. This value is consistent with the fact that titration of a fixed concentration of BA-LZ with increasing concentrations of sp-CC2 results in an increase of anisotropy (16). These results show that both the BA-CC2 and the BA-LZ monomers can bind to the sp-CC2 trimer in vitro, demonstrating a direct interaction of the CC2 and LZ coiled-coil subdomains.
Biotinylated LZ Targets in Vivo the Endogenous NEMO to Inhibit Selectively the NF-B Signaling Pathway-To determine whether NEMO is the cellular target for the most potent of our inhibitor peptides, BA-LZ, we generated wild-type and mutant LZ peptides with a substitution of BODIPY with biotin (Bio-LZ and Bio-LZ-Mut). We then performed pull-down experiments with streptavidin beads to detect a specific interaction to NEMO by Western blotting (Fig. 7A, upper panel). 70Z/3-C3 cells were incubated for 2 h without (control) or with the biotinylated peptides at 10 M concentration and then were extensively washed to remove excess peptide. This step was necessary to pull down only intracellular peptides with streptavidin beads after cell lysis. We found a specific interaction of NEMO with the Bio-LZ wild type, whereas no association was observed with the Bio-LZ mutant nor with streptavidin beads alone. This interaction was specific since the NRP protein, which displays a significant sequence homology with NEMO and shares the same "Nemo-like motif" (NLM) motif (Fig. 1B), was not able to bind to any biotinylated peptides (Fig. 7B, lower panel). The selectivity of the peptide inhibition of NF-B signaling pathway was also addressed by determining whether other signaling pathways such as mitogen-activated protein kinase (MAPK) pathway were modified by the presence of BA-LZ peptides (Fig.  7B). For this, we transiently transfected Jurkat T cells with a reporter plasmid carrying multicopies of serum-response element (SRE). The SRE-binding transcription factors, ternary complex factor (TCF) and serum-response factor (SRF), can be the targets of ERK, p38, or JNK depending on the stimulus tested (32). The stimulation of Jurkat cells with phorbol 12myristate 13-acetate and ionomycin induces a 65-fold tran-scriptional activation of the plasmid reporter pSRE-luc. This transcriptional activation was almost unchanged in the presence of 10 M BA-LZ WT or BA-LZ Mut, indicating that at least ERK and p38 pathways are not affected by our peptides.
In Vitro Mechanism of CC2 and LZ Peptides on NEMO Oligomerization-To understand the molecular mechanism for the peptide-mediated inhibition of NF-B pathway, we investigated in vitro whether sp-CC2 and sp-LZ peptides can interfere with NEMO oligomerization. We chemically synthesized the minimal oligomerization domain of NEMO lacking the His tag at its N terminus (sp-CC2-LZ) and purified from Escherichia coli the recombinant His-tagged CC2-LZ (His-CC2-LZ). His-tagged and untagged oligomers were first preincubated with a strong chaotropic salt to destroy their trimeric structure and then reassociated together to form the heterotrimer under stabilizing conditions. Once the heterotrimer was immobilized on Ni-NTA beads, the ability of sp-LZ and sp-CC2 to compete with the untagged sp-CC2-LZ was then examined by determining whether the untagged subunit is washed away upon incubation with peptides (Fig. 8). Although sp-LZ (WT) at 50 M concentration reduced markedly the amount of the untagged sp-CC2-LZ, sp-LZ-Mut did not induce any change and gave the same result as the control without peptides. When experiments were done with sp-CC2, a 500 M concentration of peptide was needed to detect a significant release of sp-CC2-LZ, indicating that the affinity for sp-CC2 is lower than that of sp-LZ. No change in the stoichiometry ratio of tagged/untagged subunits was observed when the immobilized heterotrimer was incubated without (control) or with the sp-CC2-Mut. These data demonstrate that sp-CC2 and sp-LZ peptides interfere with the oligomerization domain of NEMO.

BODIPY-free A-CC2 and A-LZ Peptides Induce Cell Death in Human Retinoblastoma
Cells-The role of the NF-B pathway in the regulation of apoptosis is well established (33,34). Treatment of human retinoblastoma Y79 cells with the SN50 peptide prevents the nuclear translocation of NF-B transcription factors, resulting in apoptosis of cancer cells (35). We investigated whether BODIPY-free but Antennapedia-conjugated (A-CC2 and A-LZ) peptides have an effect on apoptosis in human ret- inoblastoma Y79 cells (Fig. 9). BODIPY labeling was not used in this experiment as the fluorophore interferes with the cell death assay. As compared with untreated cells, cell survival was reduced by 35 and 80% in cells treated with 20 M A-CC2 and A-LZ, respectively. A similar reduction was observed when the peptide concentration was reduced to 5 M, and the incubation time of the cells with the peptide was extended to 16 h (data not shown). Treatment of the cells with the A-CC2 Mut or A-LZ Mut control peptides under the same conditions did not alter cell viability. The same was true for the Penetratin peptide alone (not shown).

DISCUSSION
It is now generally recognized that the non-catalytic protein NEMO/IKK-␥ plays a pivotal role in the activation of the NF-B transduction pathway. However, the mechanism by which this protein exerts its regulatory role on the IKK complex remains unknown. NEMO oligomerization appears to be required for its biological function (10 -12, 36). A segment called the minimal oligomerization domain, responsible for the self-association of the protein subunits, has been identified in the C-terminal half of the polypeptide chain. It consists of two coiled-coil motifs, CC2 and LZ, connected by about 15 amino acids (14,16).
In a previous study, a cell-permeable peptide corresponding to the C terminus of IKK-␣ and IKK-␤ was used to compete with the binding of NEMO, thereby preventing the assembly of the IKK complex and blocking activation of the kinases (13,37). Other cell-permeable peptides have been designed to interfere with the NF-B pathway. Horng et al. (38) used the TIRAP peptide, which competes with the TIRAP domain (Toll-interleukin-1 receptor (TIR) domain-containing adapter domain) for interaction with Toll-like receptor 4. This peptide inhibits NF-B activation in response to LPS but not in response to other proinflammatory stimuli (38). A p65 peptide that selectively inhibits NF-B activation following various proinflammatory stimuli was recently described (39). Although this peptide is a specific inhibitor of NF-B activation, its inhibitory concentration is high, thereby limiting its potential use in vivo.
Here, we used cell-permeable peptides designed to specifically interfere with the oligomerization of NEMO to block NF-B activation following treatment with LPS and possibly other stimuli. Our model is that the LZ coiled-coil can fold back onto the CC2 coiled-coil so that the oligomerization domain appears as a pseudo-hexamer made of a trimer of CC2 domains surrounded by three LZ helices (16). On the basis of this model, BA-CC2 and BA-LZ peptides were designed to interfere with the coiled-coil interfaces. We used a highly sensitive cell-based assay to monitor NF-B activation, allowing us to measure basal NF-B activity and to test very small amounts of peptides. The peptides were fused with the Antennapedia/Penetratin sequence to allow them to be transduced into cells. They were also tagged with the fluorophore BODIPY, making it possible to use FACS to monitor their internalization. We found that the NF-B pathway was strongly inhibited when pre-B lymphocytes were preincubated with BA-CC2 (IC 50 ϭ 22 M) or BA-LZ (IC 50 ϭ 3 M). The latter value is about 2 orders of magnitude lower than that of the previously described pep- tide used to disrupt the NEMO/IKK interface (13,37). Similar values were obtained when we assayed the DNA binding capacity of NF-B using the TransAM TM NF-B p65 Chemi kit provided by Active Motif Inc. (data not shown). Importantly, the peptides did not inhibit significantly basal NF-B activity, which is relevant for their possible therapeutic developments since any treatment will heavily rely on the ability of the putative drug to leave the basal NF-B activity required for cell survival undisrupted (40). The apparent small effect of BA-LZ on the basal level (Fig. 3A, inset) could be due to the fact that a very small portion of protease inhibitors can enter into cells, leading to a slight decrease on the basal level. In addition, it should be pointed out that a real basal level is in fact difficult to define precisely since experimental procedures themselves, including treatment of cells upon peptide internalization, may be sufficient to induce a slight activation of NF-B.
There are several possible explanations for the difference in the abilities of BA-CC2 and BA-LZ to inhibit the LPS-induced NF-B activation. The fact that BA-LZ was a significantly stronger inhibitor than BA-CC2 may be due to a structural difference and/or to the role played by the two coiled-coil motifs in the building of the oligomer. The dissociation/reassociation of the NEMO oligomer with BA-LZ may be faster with BA-LZ than with BA-CC2 because of increased accessibility of the LZ domain. Indeed, LZ peptides form the external ␣-helices of the CC2-LZ pseudo-hexamer (see the model in Fig. 6 of Agou et al. (16)) and are more exposed to the solvent than the trimeric central CC2 core. In addition, the N-terminal region of the LZ domain contains a so-called NLM, which is shared with four other proteins including ABIN-1 (41), ABIN-2 (42), ABIN-3/ LIND (Listeria induced) (41), and NRP (43) (Fig. 1B). Overexpression of the genes encoding ABIN-1 or ABIN-2 results in dominant negative inhibition of NF-B activation. This is essentially due to the NLM sequence as mutations in this motif abrogate the inhibitory effect (41,42). BA-LZ may thus act as a specific NF-B inhibitor by competing with NEMO for upstream activators. Interestingly, although the BA-CC2 mutant peptide had no effect, the BA-LZ mutant still inhibited the activation of the pathway by 20%. This may be due to the presence of an unaltered NLM motif within BA-LZ, which could retain some ability to compete with NEMO. BA-LZ Mut was a less potent inhibitor than BA-LZ WT, suggesting that the ␣-helical conformation of the peptides is crucial for strong inhibition. This indicates that ␣-helix-stabilizing methods combined with the binding epitope of the NLM motif could be a fruitful approach for future strategies to improve NF-B peptide inhibitors.
When discussing their mode of action, the oligomeric structure of the cell-permeable peptides must be considered. Indeed, although the free untagged peptides form trimers and dimers, the addition of a Penetratin sequence at their N terminus results in their dissociation into monomers (Fig. 5). Consistent with these data, we found that the fluorescence polarization of Penetratin-tagged peptides was not concentration-dependent up to 100 M (data not shown), showing that BA-CC2 and BA-LZ remain in a monomeric state even at high concentrations. One possible explanation is that the Penetratin motif itself adopts an ␣-helical amphipathic structure (44) that could alter the self-association of peptides by covering their hydrophobic interfaces. The fact that preservation of the hydrophobic core affects coiled-coil stability was further demonstrated by the fact that mutations in the coiled-coil interface of CC2 or LZ also led to dissociation (Fig. 5). Fluorescence polarization revealed complexes between A-free, BODIPY-free CC2, mimicking the binding site of NEMO, and Penetratin-tagged WT peptides (Fig. 5). This indicates that the hetero-association of CC2 with BA-CC2 or BA-LZ is more stable than the homo-association of BA-CC2 or BA-LZ. It thus appears that Penetratin is able to discriminate between homo-and hetero-associations by stabilizing peptide monomers.
Our proposed model for the structure of NEMO oligomerization domains is reminiscent of the fold of the gp41 ectodomain of HIV-1 (for a review, see Chan and Kim (25)). gp41-derived peptides are potent fusion inhibitors in HIV entry (25). Interestingly, so-called C-peptides mimicking the C-terminal heptad repeat 2, which are structurally related to the LZ peptide of NEMO, have the highest activity (45). Our in vitro data showed that sp-CC2 and sp-LZ peptides can disrupt the heterotrimer by competition. These properties are similar to those of N-and C-peptides of gp41, suggesting that strategies similar to those used for improving gp41-derived peptides could be extended to NEMO peptides.
As coiled-coil motifs are very common (ϳ2-4% of the amino acids in proteins are engaged in coiled-coil folds (46)), the possibility that our peptides have a biological effect due to an interaction with other coiled-coils had to be examined. It was obviously impossible to verify whether the BA-CC2 and BA-LZ peptides affected many different intracellular coiled-coils. Instead, we examined the effect of a well characterized unrelated coiled-coil peptide mimicking the leucine zipper of GCN4, which differs mainly in amino acids reported to define specificity (Fig. 4) (31, 47). The BA-GCN4 peptide had no effect on NF-B activation, consistent with the rules of coiled-coil interactions. The specificity of peptides was also addressed by demonstrating that BA-LZ targets NEMO in vivo and by showing that the MAPK signaling pathway is fully functional in the presence of NEMO-derived LZ peptides. When we also performed experiments with biotinylated CC2 peptides, no association in vivo with NEMO was detected, possibly because the affinity of the Bio-CC2 peptide for NEMO was too low.
The potential use of inhibitor peptides to block the NF-B pathway may be of particular interest given the role of this pathway in cell death and apoptosis. Indeed, constitutive activation of NF-B transcription factors is associated with most of the six essential physiological alterations that dictate the conversion of normal human cells into cancer cells (48,49). NF-B was recently shown to be involved in this process in Y79 retinoblastoma cells as cell death is more pronounced in cells treated with a peptide inhibiting NF-B translocation into the nucleus (35). Thus, our finding that the survival of Y79 retinoblastoma cells was strongly decreased after incubation with either A-CC2 or A-LZ peptides (Fig. 9) opens the possibility to extend this work in the field of anti-cancer drugs.
In conclusion, our results provide proof of concept for the development of new anti-inflammatory and anti-cancer drugs based on their potential ability to disrupt inter-and intrasubunit contacts between the coiled-coils composing the oligomerization domain of NEMO. In future work, we will search for new peptides or peptide mimetics with decreased IC 50 , as well as other organic compounds able to inhibit specifically the protein-protein interactions within the oligomer or to stabilize it in an inactive conformation.