A Novel Member of the IκB Family, Human IκB-ζ, Inhibits Transactivation of p65 and Its DNA Binding*

A novel member of the IκB family, human IκB-ζ, was identified by a differential screening approach of apoptosis-sensitive and -resistant tumor cells. The protein consists of 6 ankyrin repeats at its COOH terminus and shares about 30% identity with other IκB members. IκB-ζ associates with both the p65 and p50 subunit of NF-κB and inhibits the transcriptional activity as well as the DNA binding of the transcription factor. Interestingly, IκB-ζ is localized in the nucleus where it aggregates in matrix-associated deacetylase bodies, indicating that IκB-ζ regulates nuclear NF-κB activity rather than its nuclear translocation from the cytoplasm. IκB-ζ expression itself was regulated by NF-κB, suggesting that its activity is controlled in a negative feedback loop. Unlike classical IκB proteins, IκB-ζ was not degraded upon cell stimulation. Treatment with tumor necrosis factor-α, interleukin-1β, and lipopolysaccharide induced a strong induction of IκB-ζ transcripts. Expression of IκB-ζ was detected in different tissues including lung, liver, and in leukocytes but not in the brain. Suppression of endogenous IκB-ζ by RNA interference rendered cells more resistant to apoptosis, whereas overexpression of IκB-ζ was sufficient to induce cell death. Our results, therefore, suggest that IκB-ζ functions as an additional regulator of NF-κB activity and, hence, provides another control level for the activation of NF-κB-dependent target genes.

NF-B is an evolutionarily conserved pleiotropic transcription factor that plays a crucial role in many biological processes such as inflammation, immunity, differentiation, cell growth, tumorigenesis, and apoptosis (Refs. 2-6; for review, see Ref. 1). The mammalian NF-B/Rel family consists of RelA (p65), RelB, c-Rel, p50, and p52 that bind as homo-or heterodimers at B sites in the DNA of their target genes. This combinatorial diversity contributes to the regulation of a distinct but overlapping set of genes, in that the individual dimers have distinct preferences for different B sites that they can bind with distinguishable affinity and specificity. In addition, whether the transcription of a target gene is activated or repressed depends on the dimer combination. While RelA/ p50, RelA/c-Rel, RelB/p50, and RelB/p52 heterodimers are transcriptional activators, p50/p50 and p52/p52 homodimers generally repress transcription (7)(8)(9). Regulation of the great diversity of genes requires a precise control of NF-B, which is achieved by various mechanisms of posttranslational modification and subcellular compartmentalization as well as by interactions with other cofactors or corepressors. Dysregulation of NF-B activation results in a wide range of human disorders including inflammatory and neurodegenerative diseases (10,11) and different types of cancer (4,12).
NF-B activity is also tightly regulated by interaction with IB proteins (13). Characteristic for all IB proteins is a domain of multiple ankyrin repeats that bind to the conserved Rel homology domain of NF-B proteins. Various members of the IB family target different NF-B complexes, e.g. IB-␣ and IB-␤ interact preferentially with p65/ p50 and c-Rel/p50 heterodimers, whereas IB-⑀ binds only to p65 and c-Rel hetero-and homodimers (14,15), and Bcl-3 associates exclusively with p50 or p52 homodimers (16). Even if different IB proteins interact with the same NF-B dimers, the transcription factor can be regulated differentially due to functional differences between the IB proteins. For instance, upon cell stimulation IB-␣ is rapidly degraded, leading to an immediate but transient activation of NF-B. In contrast, IB-␤ persists over a longer time, and its degradation causes a more delayed and sustained activation of the transcription factor (14,17).
In most cell types NF-B is sequestered in the cytoplasm as an inactive complex bound to IB proteins. A variety of stimuli, including tumor necrosis factor (TNF), 3 interleukin-1 (IL-1), and lipopolysaccharide (LPS), lead to the phosphorylation of IBs by the IB kinase complex, which promotes their ubiquitination and proteasomal degradation (18). This event exposes a nuclear localization signal of NF-B leading to its nuclear translocation and activation of target genes. The rapidly resynthesized IB-␣, which itself is an NF-B target gene, can enter the nucleus, dissociate NF-B from the DNA, and translocate it back to the cytoplasm, thereby terminating gene transcription (19). In contrast to the typical IB proteins (IB-␣, IB-␤, IB-⑀) that are preferentially localized in the cytoplasm and behave exclusively as inhibitors, Bcl-3 is a nuclear protein that preferentially promotes B-dependent gene transcription. Bcl-3 can cause derepression of transcription by removing p50 and p52 dimers, which are transcriptionally inactive, from B sites (20). Alternatively, Bcl-3 can lead to direct transcriptional activation by forming a ternary complex with DNA through these homodimers (21)(22)(23).
In a continuous attempt to define molecular mechanisms involved in apoptosis, we performed a suppression subtractive hybridization approach with two HeLa cell lines that differ in their susceptibility to TNF (24,25). Among several differentially expressed genes, a cDNA clone was identified encoding a novel IB protein. In this study, we describe the structural and functional characterization of this protein, termed IB-. Unlike typical IB proteins, IB-is not rapidly degraded but, rather, stably accumulates in the nucleus where it inhibits NF-B activity. Interestingly, this inhibitory capability is opposite to the transactivating function of the nuclear Bcl-3 protein. Furthermore, IBis only inducibly expressed in different cell types. Thus, these features of IB-, which are distinct from the cytosolic IBs and Bcl-3, may confer an additional control mechanism of NF-B activity.

MATERIALS AND METHODS
Reagents and Antibodies-Moloney murine leukemia virus reverse transcriptase, Lipofectamine/PLUS TM reagent, and the anti-c-Myc monoclonal antibody were from Invitrogen. Taq DNA polymerase and dNTPs were from Eppendorf (Hamburg, Germany), luciferin was from Applichem (Darmstadt, Germany), and o-nitrophenyl-␤-D-galactopyranoside was from Calbiochem. TNF-␣, LPS, anti-actin, and anti-FLAG antibodies, and the protease inhibitors phenylmethylsulfonyl fluoride, aprotinin, leupeptin, and pepstatin were obtained from Sigma. IB-␣ polyclonal antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA), and anti-CD95 monoclonal antibody was from BioCheck (Münster, Germany). The mouse monoclonal antibody against human IB-was generated in our own laboratory. Hexamer nucleotides were from Roche Applied Science. Horseradish peroxidaselabeled goat anti-mouse IgG and anti-rabbit IgG were from Bio-Rad, RNAguard and protein G-Sepharose were from Amersham Biosciences, IL-1␤, BAY 11-7082, and MG-132 were from Biomol (Hamburg, Germany), and Mitotracker Red was from Molecular Probes (Leiden, The Netherlands). The pSilencer small interfering RNA (siRNA) expression vector kit was from Ambion (Huntingdon, Cambridgeshire, UK), and the Human RNA Master and MTN Blot were from Clontech (Palo Alto, CA).
RNA Extraction-Total RNA was extracted with TRI reagent (Sigma) and analyzed for integrity by electrophoresis on a formaldehyde agarose gel stained with ethidium bromide.
Suppression Subtractive Hybridization-TNF-resistant H21 and TNFsensitive D98 cells were stimulated with a combination of TNF (100 pg/ml) and cycloheximide (10 g/ml) for 1 h, and the poly(A) ϩ RNA was isolated from the total RNA using an oligo(dT)-cellulose column (Amersham Biosciences). The suppression subtractive hybridization was performed with the PCR-Select cDNA subtraction kit (Clontech) as described previously (27,28).
Rapid Amplification of cDNA Ends (RACE); Plasmids and Retroviral Constructs-The initiation codon of the 5Ј-untranslated region of human IBwas determined by 5Ј-rapid amplification of cDNA ends (RACE) using the SMART RACE cDNA amplification kit (Clontech) according to the manufacturer's instructions. The full-length cDNA was obtained by polymerase chain reaction and subcloned into the pcDNA4/Myc-His (Invitrogen), pCMV-Tag4 (Stratagene), pGEX-2T (Amersham Biosciences), and pEGFP-C1 or pCFP-C1 (Clontech) expression plasmids. Both the NH 2 -terminal region (amino acids 1-317) and the COOH-terminal region (amino acids 318 -618) as well as NF-B subunits p65 and p50 were subcloned into the pcDNA4/Myc-His plasmid. The correct cDNA sequences were confirmed by sequence analyses. IBwas also cloned into the bicistronic retroviral vector pCFG5-IEGZ, allowing expression of GFP from the same mRNA as IB-. A transdominant non-degradable IB-␣ (S/A) mutant cloned in pCFG5-IEGZ was provided by S. Ludwig. Retrovirus generation and infection of target cells were performed essentially as described (29).
Northern and Dot Blot Hybridization-To detect expression of IB-in normal human tissues, the commercially available MTN and RNA Master Blots (Clontech) were used. A 714-bp BamHI/Hind III fragment was taken as probe. Hybridization with the ␣-32 P-labeled cDNA probe and washing procedure was carried out according to standard protocols.
Luciferase Reporter Gene Assay-To determine NF-B-dependent transcriptional activity, a NF-B luciferase reporter plasmid containing six B binding sites in front of the luciferase promoter was used. A ␤-galactosidase (lacZ) vector served as an internal control for transfection efficiency. HEK 293 cells were seeded into 6-well plates 1 day before transfection. The cells were transiently transfected with the NF-B luciferase plasmid and the lacZ plasmid together with the indicated expression plasmids using the Lipofectamine/PLUS reagent. 24 -48 h after transfection, cells were either left unstimulated or were stimulated with TNF or IL-1 for 6 h, lysed in 25 mM glycylglycine, pH 7.8, 15 mM MgSO 4 , 4 mM EGTA, 1 mM dithiothreitol, 1% Triton X-100, and centrifuged at 13,000 rpm at 4°C for 5 min. 10 l of the supernatant were assayed in 100 l of assay buffer (15 mM potassium P i , pH 7.8, 25 mM glycylglycine, 15 mM MgSO 4 , 4 mM EGTA, 1 mM dithiothreitol, 2 mM ATP) using a microplate luminometer (Berthold Technologies, Bad Wildbad, Germany). Light emission was measured after injection of 100 l of luciferin (0.3 mg/ml), and the values were normalized to ␤-galactosidase activity.
Electrophoretic Mobility Shift Assay-HEK 293 cells were plated at 3 ϫ 10 5 /well in 6-well plates 1 day before transfection with 500 ng of the p65 or p50 plasmid and increasing amounts (1-16 g) of the IB-or, as a control, the IB-␣ plasmid. The total amount of transfected DNA was normalized with the empty vector control. Cells were lysed 48 h after transfection, and the binding reaction was performed for 30 min at room temperature in a 20-l volume containing 4 l of extract, 4 l of 5ϫ binding buffer (20 mM HEPES, pH 7.5, 50 mM KCl, 2.5 mM MgCl 2 , 1 mM dithiothreitol, 20% Ficoll), 2 g of poly(dI-dC), 2 g of bovine serum albumin, and the 32 P-labeled NF-B oligonucleotide. Samples were loaded on a 4% non-denaturing polyacrylamide gel and run in 0.5ϫ Tris borate EDTA buffer, pH 8. After drying, the gels were exposed to x-ray films.
Preparation of Whole, Cytosolic, and Nuclear Cell Extracts-To obtain whole protein extracts, cells were lysed in buffer containing 150 mM NaCl, 50 mM Tris, pH 7.4, 1 mM dithiothreitol, 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml concentrations each of aprotinin, leupeptin, and pepstatin for 30 min on ice and then centrifuged at 13,000 rpm for 30 min at 4°C. For the preparation of cytosolic and nuclear extracts, HeLa D98 cells were collected by centrifugation and broken by 4 freeze-thaw cycles in 10 mM HEPES, pH 7.9, 5 mM MgCl 2 , 0.25 M sucrose, 5 mM NaF, 10 mM ␤-mercaptoethanol, 0.1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml each of aprotinin, leupeptin, and pepstatin. After incubation for 10 min on ice and a centrifugation step (14,000 rpm, 5 min, 4°C), the supernatant (cytosolic fraction) was transferred to a fresh tube. The pellet was resuspended in high salt buffer (400 mM NaCl, 10 mM HEPES, pH 7.9, 1.5 mM MgCl 2 , 0.1 mM EDTA, 25% glycerol, 5 mM NaF, 10 mM ␤-mercaptoethanol) containing protease inhibitors and incubated for 30 min on ice. Thereafter, the freeze-thaw cycle was repeated four times, and the nuclear extract (supernatant, soluble fraction) as well as the nuclear pellet (insoluble fraction) was collected by centrifugation. Protein concentrations were determined with the Bio-Rad protein assay.
SDS-PAGE and Western Blotting-Equal amounts of the proteins were separated in a 10% SDS-polyacrylamide gel and transferred elec-trophoretically to a polyvinylidene difluoride membrane (Amersham Biosciences). The membrane was blocked with 5% bovine serum albumin in phosphate-buffered saline, 0.2% Tween for 2 h, and incubated overnight at 4°C with the IB-, IB-␣, poly(ADP-ribose) polymerase, caspase-3, or actin antibodies in blocking buffer. After incubation for 1 h with the secondary antibody (horseradish peroxidase-labeled goat antimouse IgG or anti-rabbit IgG) in phosphate-buffered saline, 0.2% Tween, 5% bovine serum albumin, the proteins were visualized by enhanced chemiluminescent staining using ECL reagents (Amersham Biosciences).
Immunoprecipitation-HEK 293 cells were transiently transfected with FLAG-tagged IB-or IB-␣ together with Myc-tagged p65 or p50 for 24 h using Lipofectamine/PLUS TM reagent. Then lysis was performed for 30 min on ice in 20 mM HEPES, pH 7.4, 84 mM KCl, 10 mM MgCl 2 , 0.2 mM EDTA, 0.2 mM EGTA, 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml each of aprotinin, leupeptin, and pepstatin. After centrifugation the supernatants were incubated with protein G-Sepharose and either anti-FLAG or anti-IB-␣ antibody overnight at 4°C. The immunoprecipitates were resolved by 10% SDS-PAGE, transferred to a polyvinylidene difluoride membrane, blocked, and incubated overnight at 4°C with a monoclonal anti-c-Myc antibody (Invitrogen, 1:5000). A horseradish peroxidase-labeled goat anti-mouse IgG served as a secondary antibody.
Immunofluorescence Microscopy-Cells were seeded on coverslips and transfected with the indicated plasmids. FLAG-tagged silencing mediator of retinoid and thyroid hormone receptor (SMRT) and yellow fluorescent protein-tagged histone deacetylase-5 (HDAC-5) constructs were obtained from Dr. R. M. Evans. After 18 h of incubation, the cells were fixed for 5 min with 3.7% formaldehyde in phosphate-buffered saline and incubated in IF buffer (4% bovine serum albumin, 0.05% saponin in phosphate-buffered saline). Cells were then labeled with primary antibodies (anti-PML, Santa Cruz Biotechnology; anti-FLAG, Sigma) in IF buffer at 4°C overnight. After washing, incubation with secondary antibodies (Alexa-594-coupled chicken anti-mouse IgG, Molecular Probes) was performed for 4 h. Finally, coverslips were mounted on glass slides using DAKO fluorescence mounting medium supplemented with Hoechst 33342 (500 ng/ml). Digital pictures were captured using a Leica TCS-SP2/AOBS confocal microscope.
Construction of Small Interfering RNAs and Stable Transfection-For the suppression of IB-, two different siRNAs were designed using the Dharmacon siDESIGN Center. The selected sense sequences were 5Ј-GATTCGTTGTCTGATGGAC-3Ј and 5Ј-TACTCGGAACTTG-GAGAAC-3Ј. Oligonucleotides containing the sense, loop, antisense sequence, and a polythymidine tract were annealed and ligated into the pSilencer siRNA expression vector according to the manufacturer's instructions (Ambion). HeLa D98 cells were stably transfected by electroporation using the Bio-Rad gene pulser (500 microfarads, 160 V). After hygromycin selection, several clones were obtained, and the successful reduction of IB-expression was controlled by Western blot analysis.
Cytotoxicity Assay-Parental HeLa D98 cells and three different clones stably expressing IB-siRNA were seeded into 96-well microtiter plates at a density of 3-4 ϫ 10 5 /ml. Before exposure to various concentrations of the apoptosis-inducing reagents TNF and anti-CD95, each combined with cycloheximide (10 g/ml), the cells were stimulated with IL-1 for 2 h. For determination of cell death, the standard crystal violet assay was employed (31). The viable cells were stained with 20% methanol containing 0.5% crystal violet and solubilized in 33% acetic acid. The absorbance was measured at A 560 nm . Percentage-specific cell death is defined as 100 Ϫ (A 560 of test well ϫ 100/A 560 of untreated well). Each experiment was performed independently at least three times, and an individual experiment was carried out in triplicate.

Isolation of a cDNA Encoding a Novel Member of the IB Family-To
identify genes involved in the modulation of apoptosis, the human HeLa cell lines D98 (TNF-sensitive) and H21 (TNF-resistant) were exposed to TNF combined with cycloheximide for 1 h, and the mRNAs were subjected to the suppression subtractive hybridization procedure. Among several differentially expressed genes, a cDNA clone was identified encoding a protein that contains six ankyrin repeats at its COOH terminus sharing about 30% identity with different members of the IB protein family. Northern blot analysis revealed that the mRNA has a length of about 3.5 kilobases kb (data not shown). The full-length cDNA contains an open reading frame of 1857 bp encoding a protein of 618 amino acids. The sequence data were deposited in the data base (Gen-Bank TM DQ224339). The nucleotide and amino acid sequences are shown in Fig. 1, A and B, respectively. Besides the COOH-terminal ankyrin repeats, the gene product consists of a potential PEST domain which, however, in contrast to IB-␣ and IB-␤, is located at the NH 2 terminus (amino acids 84 -104). The other NH 2 -terminal regions exhibit no significant homology with other proteins.
The structural similarity to IB members suggested that we had identified a novel IB protein. Data base analysis revealed an 80% homology to a mouse protein, termed molecule possessing ankyrin repeats induced by lipopolysaccharide (MAIL). This protein exists in two isoforms, MAIL-L and MAIL-S, which is identical to IB- (32,33). During the course of our functional characterization of the novel gene product, sequences of different human MAIL isoforms were deposited in the data base. Our identified gene product is most similar to MAIL-S (Gen-Bank TM AF548362), with the difference that it lacks exon 2 (nucleotides 193-268) in the 5Ј-untranslated region and therefore represents a to date unknown isoform of MAIL. Because the MAIL proteins are also known as nuclear factor of light polypeptide gene enhancer in B-cells inhibitor (NFKBIZ), transcript variant 1 and transcript variant 2, we termed the novel isoform human IB-, transcript variant 3. In the following this variant is referred to as IB-.
IB-Is Expressed in Different Human Tissues-To obtain an mRNA expression profile of IB-, a commercially available RNA dot blot was used. Expression was detectable in different human tissues such as lung, placenta, and liver and also in peripheral blood leukocytes. In contrast, IB-was not expressed in different regions of the brain (Fig. 1C). To verify these results, a Northern blot was performed with mRNA from 12 different human tissues. Again, lung and peripheral blood leukocytes showed the highest expression, liver and placenta showed moderate expression, and spleen, kidney, skeletal muscle, and heart showed a low expression level, whereas no expression could be detected in the brain (Fig. 1D).

NF-B Promoter Activity Is Inhibited by the COOH-terminal Domain of IB--
The presence of ankyrin repeats in the COOH-terminal region of IB-suggested that we had identified a bona fide NF-B inhibitor, which encouraged us to investigate its effect on the regulation of NF-B activity. To this end, HEK 293 cells were transiently transfected with a B-dependent luciferase reporter plasmid together with a ␤-galactosidase vector and a plasmid containing the cDNA for either IB-or as a control IB-␣. As measured in a luciferase reporter gene assay, activation of NF-B was observed when cells were exposed to TNF (Fig. 2A). IB-was able to inhibit NF-B activity in TNF-stimulated as well as in unstimulated cells in a concentration-dependent manner, indicating that IB-indeed belongs to the class of proteins that inhibit NF-B. Compared with TNF, the cytokine IL-1 was a weaker NF-B activator in HEK 293 cells. Nevertheless, the IL-1-induced NF-B activation was also reduced by IB- (Fig. 2B), confirming its inhibitory effect on NF-B. To find out which functional domain of IB-is responsible for this inhibition, truncated mutants of IB-were constructed. The COOH-terminal region containing the ankyrin repeats efficiently inhibited NF-B activity to an even greater extent than fulllength IB-, whereas the NH 2 -terminal domain had no effect (Fig. 2C). As a control, in all experiments the effect of IB-␣ was investigated in parallel. IB-␣ was a much stronger NF-B inhibitor, indicating that the transcriptional activity of NF-B is regulated differentially by these two IB proteins. Such a weaker inhibition of NF-B activity in comparison to IB-␣ has been also reported for IB-⑀ (34).
IB-Binds to the p65 and p50 Subunits of NF-B and Inhibits Its DNA Binding and Transactivation Activity-Because IB-was identified as an NF-B inhibitor, the question arose of which NF-B subunits associate with IB-. Therefore, HEK 293 cells were transiently transfected with IB-and p65 or p50, and the co-immunoprecipitated lysates were analyzed by Western blotting. IB-was found to interact with both the p65 and the p50 subunit (Fig. 3A). As expected, a strong association of p65 and p50 with IB-␣ was observed.
The p65/p50-IB-interaction suggested that IB-might influence the binding activity of NF-B to the DNA. Indeed, gel-shift analyses revealed that IB-inhibited the DNA binding of p65/p50, since the p65 as well as the p50 band disappeared gradually with increasing amounts of IB- (Fig. 3B). Compared with IB-␣, a higher amount of IB-was necessary to obtain the same effect, demonstrating that IB-␣ functions as a much more efficient inhibitor of NF-B DNA binding than IB-.
Next, we investigated whether the binding of IB-to p65 affected its transactivation activity. Ectopic expression of p65 caused a strong activation of NF-B that was substantially inhibited by IB-and almost completely blocked by IB-␣ (Fig. 3C). Interestingly, although both IB proteins associated with the same NF-B subunits, they differentially regulated DNA binding activity of NF-B and transactivation of p65.
IB-Is Inducible at the mRNA and Protein Level-For further characterization of IB-, its regulation was investigated. HeLa D98 cells were exposed to TNF and IL-1, and the mRNA expression was analyzed in comparison to IB-␣. Very low levels of IB-mRNA were observed in unstimulated cells (Fig. 4A). TNF and IL-1 were able to induce IB-mRNA in a time-dependent manner. Maximal expression of the IB-mRNA occurred within 30 -60 min after stimulation. Although a slight decline was observed afterward, a relatively high expression level persisted over the whole time course of the experiment. A similar induction profile was detected for transcription of IB-␣ (Fig. 4A). The stimulatory effect of the different agents was not restricted to the HeLa cells, as MCF-7 breast cancer cells were similarly affected (Fig. 4B). Because the IL-1 and the LPS receptor activate similar intracellular signaling pathways, it could be assumed that human IB-is also inducible by LPS. Because neither the HeLa cells nor the MCF-7 cells responded toward LPS, the human monocytic cell line MonoMac6 was treated with LPS. In fact, also LPS led to an increased IB-expression with a time course similar to that observed for TNF and IL-1 (Fig. 4C).
In the next set of experiments we examined the regulation of IBprotein expression using a monoclonal antibody that we had generated against the recombinant protein. In quiescent MCF-7 cells IB-was not expressed, but after stimulation with IL-1 a maximal induction of IB-protein expression occurred within 1-2 h and declined thereafter (Fig. 5A, left panel). In contrast, preexisting IB-␣ was degraded rapidly within 15-30 min and resynthesized 1 h after cytokine stimulation. IL-1-stimulated IB-expression showed a similar induction pattern in HeLa cells (Fig. 5A, middle panel). A weaker but clearly significant induction of IB-protein was also seen after stimulation of HeLa cells with TNF (Fig. 5A, right panel). Furthermore, a rapid induction of IBprotein was found in different IL-1-stimulated cell types, such as the human lung carcinoma cell line A549, the hepatocellular carcinoma cell line HepG2, and human HaCaT keratinocytes (Fig. 5B).
As for the mRNA expression studies, MonoMac6 cells were used to examine the effect of LPS on IB-protein expression. Indeed, the IBprotein level increased dramatically starting 1 h after exposure to LPS and persisted over at least 24 h (Fig. 5C). It is noteworthy that in LPSstimulated monocytes we reproducibly observed a biphasic induction of IB-protein expression. Furthermore, in addition to the 85-kDa band of IB-, in these cells a protein of ϳ70 kDa was induced that presumably corresponds to the short isoform of IB-.
IB-Is Regulated by NF-B-After stimulation of NF-B, both the mRNA of IB-and IB-␣ were up-regulated. Therefore, we asked whether this transcription factor regulates IB-in a similar manner as known for IB-␣ (35,36). To this end, HeLa D98 cells were incubated for 1 h before the addition of IL-1 with the proteasome inhibitor MG-132, which prevents the degradation of IB-␣ and, thus, NF-B activation. Analysis of protein expression revealed that, as expected, IL-1 treatment induced a strong degradation of IB-␣ that was almost completely prevented in the presence of MG-132 (Fig. 6). Interestingly, the IL-1-induced IB-expression was markedly reduced by MG-132, indicating that NF-B activation might be necessary for the induction of IB- (Fig. 6, upper left panel). This assumption was supported at the transcriptional level, since the IL-1-induced mRNA expression levels of both, IB-␣ and IB-, decreased when cells were pretreated with MG-132 (Fig. 6, upper right panel). These results were further confirmed by using the IB kinase inhibitor BAY 11-7082 that prevents the phosphorylation of IB-␣ and, hence, NF-B activation. IL-1-induced IB-expression was diminished dose-dependently at the mRNA and protein level by BAY 11-7082 (Fig. 6, lower panels). Thus, these data show that IB-is regulated, at least in part, by NF-B.
IB-Localizes to Nuclear Matrix-associated Deacetylase Bodies-To determine the intracellular localization of IB-, cytoplasmic and nuclear fractions were isolated from HeLa D98 cells and analyzed by Western blotting. IB-was preferentially expressed in the nuclear fractions (Fig. 7A). Under basal conditions a very low expression of IBwas detected that was strongly induced after a 2-4-h treatment with IL-1 (Fig. 7A). As a control for the purity of the subcellular fractions, the membrane was reblotted with anti-poly(ADP-ribose) polymerase and anti-caspase-3 antibodies as nuclear and cytoplasmic markers, respectively. Additionally, the localization was examined by immunofluorescence microscopy in cells transiently transfected with GFP-tagged  IB-. Interestingly, these analyses revealed that IB-was not evenly distributed in the nucleus but aggregated in distinct dot-like structures (Fig. 7B).
To identify the nature of these nuclear bodies, we performed detailed colocalization studies using different marker proteins. IBneither colocalized with promyelocytic leukemia (PML) protein (Fig. 8, right  panel) nor with SC-35 (data not shown), suggesting that it was not associated with PML or RNA splicing bodies. However, the localization of IBcompletely coincided with the nuclear corepressor SMRT (Fig.  8, left panel), which has been found to be retained in the recently identified matrix-associated deacetylase bodies (37). Indeed, also HDAC5 was localized in same subnuclear structures (Fig. 8, middle panel). The results, therefore, suggest that IBis a nuclear IB protein that might function through modulating HDAC activity.
IB-Renders Cells More Sensitive to Cell Death-Because we had originally identified IB-as a TNF-inducible gene product that was overexpressed in apoptosis-sensitive but not in apoptosis-resistant HeLa cells, the functional role of IB-in the regulation of cell death was examined. RNA interference was used to suppress the IL-1-induced expression of IB-. HeLa D98 cells were stably transfected with the pSilencer vector containing different siRNAs for IB-. Western blot analysis confirmed that, compared with the wild-type cells, the IL-1induced IB-protein expression was strongly or even almost completely inhibited in the hygromycin-selected siRNA clones (Fig. 9A). The IB-knock-down and wild-type cells were then either left unstimulated or pretreated with IL-1 for 2 h before the exposure to various concen-FIGURE 6. IB-is regulated by NF-B. Western blot (left panels) and RT-PCR analyses (right panels) for the expression of IB-and IB-␣ in HeLa D98 cells after stimulation with IL-1 for the indicated times. Cells were pretreated 1 h before stimulation with proteasome inhibitor MG-132 or the IB kinase inhibitor BAY 11-7082. As a loading control for the RT-PCR and Western blot analyses, the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and actin expression, respectively, were examined.  trations of TNF or anti-CD95. As determined by the crystal violet cytotoxicity assay, the different apoptosis-promoting stimuli led to a pronounced and dose-dependent induction of HeLa wild-type cell death. Interestingly, compared with the wild-type cells, all IB-siRNA clones used in this assay were more resistant to apoptosis induced by the various stimuli (Fig. 9B).
To further substantiate the results obtained by RNA interferencemediated knockdown of IB-, we overexpressed IB-by retroviral infection of HT-1080 cells. We employed retroviral constructs in which expression of either IB-or IB-␣ was coupled to GFP expression through an internal ribosome entry site. After 2 days of incubation, expression of IB-resulted in pronounced cytotoxicity, as evidenced by a strong cytoplasmic and nuclear condensation and detachment of the GFP-positive cells (Fig. 9C). In contrast, cells expressing only GFP remained mostly viably. In several experiments we noticed that the cytotoxic effect of IB-appeared even stronger than that observed after expression of IB-␣. Thus, these results indicate that IB-promotes apoptosis.

DISCUSSION
Using the suppression subtractive hybridization technique, we have identified a gene product that was differentially expressed in apoptosissensitive and -resistant HeLa cells. The gene encodes a novel IB protein with six ankyrin repeats and a potential NH 2 -terminal PEST domain. Sequence comparison revealed that it is most similar to human MAIL-S (GenBank TM AF548362) or nuclear factor of light polypeptide gene enhancer in B-cells inhibitor (NFKBIZ), transcript variant 2 (GenBank TM NM_001005474), which has, however, not been characterized so far. Based on this information, we termed our newly identified clone IB-, transcript variant 3. Although both MAIL-S, which had been deposited in the data base during this study, and our gene product encode a protein of 618 amino acids, the lack of exon 2 in the 5Ј-untranslated region of IB-may have regulatory consequences.
IB proteins interact via their ankyrin repeats with the Rel homology domain of NF-B. The transcriptional activity of NF-B was inhibited by full-length IB-and even more by a COOH-terminal construct containing the ankyrin domain. Thus, it seems that elements in the amino terminus can influence the effect of IB-directly or that the structure of the carboxyl terminus is more favorable to interact with NF-B subunits.
Control of NF-B activation is partially regulated by preferential association of specific NF-B dimers with certain members of the IB family. IB-␣ and IB-␤ interact with p65/p50 and c-Rel/p50 heterodimers, IB-⑀ associates only with homo-or heterodimeric complexes containing p65 and c-Rel (14,15,34), and Bcl-3 interacts specifically with p50 and p52 homodimers (22). In contrast, p100 and p105, which are precursors of the Rel proteins p52 and p50, respectively, appear to bind efficiently to all Rel proteins. Moreover, functional differences between various IB proteins allow the binding to the same NF-B dimer. For example, the classical NF-B heterodimer p65/p50 interacts with IB-␣ and IB-␤, but both IB proteins show differences in their response to various NF-B inducers and in their kinetics of degradation (14,38). In our study we were able to demonstrate that IB-also associates with p65 and p50 but was less effective in inhibiting NF-B transcriptional activity as compared with IB-␣. In addition, a higher amount of IB-was necessary to inhibit the binding of NF-B to the DNA, possibly due to a higher affinity of p65/p50 to IB-␣. A considerable weaker inhibition of NF-B DNA binding compared with IB-␣ has been reported for IB-⑀ (34).
Unlike the classical IB proteins, our study shows that the novel IBprotein is not controlled by inducible degradation. Furthermore, IBwas not expressed in quiescent cells but was strongly induced upon stimulation, suggesting that different mechanisms take place in the regulation of NF-B by these inhibitors. Because IB-is localized in the nucleus, it is likely that it controls the activated NF-B in this compartment rather than the translocation of NF-B into the nucleus. Nevertheless, the remarkable up-regulation of IB-after NF-B activation and its down-regulation by BAY and MG-132 together with the identification of IB-as NF-B inhibitor suggest that this IB protein is regulated by NF-B in a negative feedback loop. Negative autoregulatory loops provide an effective mechanism for controlling NF-B activity, especially because fast and slow negative feedback loops exist. In contrast to the rapid IB-␣ expression (39), IB-⑀ represents a slower negative feedback regulatory mechanism (15). Because the mRNAs of IB-and IB-␣ were induced with similar kinetics, our data suggest that IB-represents an additional fast negative feedback loop for the control of NF-B activation.
Our results indicate that there might exist species-specific differences in the regulation of NF-B by IB-. Although human and mouse IB-(also called MAIL-S or INAP) are highly homologous at their COOH terminus, both proteins differ particularly at their NH 2 terminus. The mouse homologue has been demonstrated to interact only with the p50 subunit of NF-B but not with p65 (33). Another study (40) even found no interaction of mouse IB-with p50, p52, or p65. Therefore, human IB-cannot only regulate the DNA binding of NF-B but also its transcriptional activity through association with the NF-B subunit p65. Moreover, a different gene expression profile might occur in human and mouse systems due to the different composition of the NF-B dimers.
Consistent with our data, mouse IB-is induced in response to LPS and IL-1 (33). The cytoplasmic domains of the Toll-like receptors and the IL-1 receptor are homologous and, hence, activate the same signaling pathways leading to the activation of NF-B (41,42), which then can induce IB-in an autoregulatory manner. It was recently shown that induction of mouse IB-through Toll-like/IL-1 receptors was specifically due to mRNA stabilization (43). In contrast, our data also demonstrate that the induction of human IB-is exerted additionally via the TNF receptor pathway. Whereas TNF had no effect on the induction of mouse IB-(33), the human homologue was TNF-responsive. Whether the differential inducibility of human and mouse IB-reflects a species-or a cell type-specific difference remains to be shown. In this context, it has been reported that NF-B activity was required but not sufficient for the induction of mouse IB- (44). Our data also suggest that human IB-is regulated by NF-B. Whether NF-B activity is sufficient to induce human IB-is under further investigation.
Besides its function as a negative regulator of NF-B activation, it was recently demonstrated that mouse IB-exerts transcriptional activity. Analyses of IB--deficient mice revealed that it was essential for the expression of numerous LPS-inducible genes including IL-12 p40, GM-CSF (granulocyte-macrophage colony-stimulating factor), G-CSF (granu-locyte CSF), C/EBP-␦, and endothelin (45). Moreover, the Toll-like receptor/IL-1 receptor-mediated production of IL-6 was profoundly inhibited in IB--deficient cells. In this context it has been reported that the IL-6 promoter was activated by mouse IB-in the macrophage cell line RAW 264.7. Interestingly, whereas IB-enhanced LPS-mediated IL-6 promoter activity (45), IBNS, another related nuclear IB protein, suppressed IL-6 gene transcription (46). Transcriptional activation of NF-B target genes has been also shown for Bcl-3 that like IB-is not degraded by the proteasome and also localized in the nucleus. Bcl-3 can cause DNA-bound p50 homodimers to dissociate from the B site, permitting these inhibiting NF-B species to be replaced by the transactivating members p65, RelB, or c-Rel. Bcl-3 can also form a ternary complex with DNA-bound p50 or p52 homodimers and activate transcription directly, an activity that requires both NH 2 -and COOH-terminal domains of Bcl-3 (21,22,47). To date, only the cyclin D1 promoter has been shown to be directly activated by Bcl-3-p52 complexes (48,49). A possible function of human IB-as a transcriptional activator has not been investigated so far. However, recently it has been suggested that the human MAIL isoforms might harbor a transactivation domain in their NH 2 terminus (50). Therefore, it is conceivable that human IB-may directly activate the transcription of target genes. Nevertheless, if so, the subset of target genes affected and the molecular mechanism for their induction might be different from mouse IB-due to the TNF-inducibility of human IB-and its association with the p65/p50 heterodimer.
In line with its function as a negative regulator of NF-B, we found a proapoptotic effect of human IB-. Suppression of endogenous IBrendered cells more resistant to apoptosis induced by TNF and anti-CD95, whereas overexpression of IB-was sufficient to induce cell death. Although the molecular mechanisms of modulation of cell survival by IB-are currently unknown, these results are consistent with the anti-apoptotic function of NF-B. Products of NF-B target genes that inhibit apoptosis include members of the Bcl-2 family (Bcl-x L , Bfl-1/A1), inhibitors of apoptosis proteins (c-IAP1, c-IAP2, XIAP), adaptor molecules (TNF receptor-associated factor (TRAF)1, TRAF2), and FLICE inhibitory protein (c-FLIP) (for review, see Ref. 5).
It has become increasingly clear that the activation of NF-B is not only controlled in the cytoplasm but, presumably even more importantly, also modulated in the nucleus. Not only phosphorylation and acetylation of NF-B itself play a critical role in the regulation of its transcriptional activity (51,52) but also interactions with other nuclear proteins including histone deacetylases, coactivators, and other transcription factors (9,(53)(54)(55)(56). Our finding that IB-colocalizes with HDAC5 and the corepressor SMRT in matrix-associated deacetylase nuclear bodies is, therefore, highly intriguing. Although further functional studies in this respect are required, the localization of IB-in these subnuclear structures points to role of IB-in modulation of HDAC activity and chromatin structure. Our study together with the results of other groups, thus, suggests that nuclear IB proteins add an additional layer to the already complex regulatory mechanisms that activate or repress NF-B targets genes. Although the molecular mechanism of NF-B regulation by human IB-has to be elucidated, this IB protein might play an important role as regulator of NF-B activity, especially because it is, to our knowledge, the first nuclear IB protein that binds to the p65 and p50 subunit of NF-B and regulates the transcription factor in a negative way.