Phorbol Esters and Cytokines Regulate the Expression of theNEMO-related Protein, a Molecule Involved in a NF-κB-independent Pathway*

The NF-κB signaling pathway plays a crucial role in the immune, inflammatory, and apoptotic responses. Recently, we identified the NF-κB Essential Modulator (NEMO) as an essential component of this pathway. NEMO is a structural and regulatory subunit of the high molecular kinase complex (IKK) responsible for the phosphorylation of NF-κB inhibitors. Data base searching led to the isolation of a cDNA encoding a protein we called NRP (NEMO-related protein), which shows a strong homology to NEMO. Here we show that NRP is present in a novel high molecular weight complex, that contains none of the known members of the IKK complex. Consistently, we could not observe any effect of NRP on NF-κB signaling. Nonetheless, we could demonstrate that treatment with phorbol esters induces NRP phosphorylation and decreases its half-life. This phosphorylation event could only be inhibited by K-252a and stauroporin. We also show that de novo expression of NRP can be induced by interferon and tumor necrosis factor α and that these two stimuli have a synergistic effect on NRP expression. In addition, we observed that endogenous NRP is associated with the Golgi apparatus. Analogous to NEMO, we find that NRP is associated in a complex with two kinases, suggesting that NRP could play a similar role in another signaling pathway.

The transcription factor NF-B plays a pivotal role in many cellular processes such as immune responses, inflammation, and apoptosis (1,2). NF-B is composed of homo-and heterodimers of various members of the NF-B/Rel family (3,4) and is retained in an inactive form in the cytoplasm by an inhibitory protein belonging to the IB family, mainly represented by IB␣, IB␤, and IB⑀ (5)(6)(7).
In response to diverse stimuli, including IL-1, 1 LPS, TNF␣, or PMA, as well as several viral proteins, active NF-B trans-locates to the nucleus as a result of the complete proteolytic degradation of the IB proteins. This mechanism has been best studied for the inhibitor IB␣ and demonstrated to involve phosphorylation on two specific serine residues (8 -13) followed by polyubiquitination and degradation by the 26 S proteasome (14). More recently a specific serine protein kinase activity responsible for IB␣ phosphorylation has been identified as a large cytoplasmic complex (600 -800 kDa) containing two catalytic subunits (IKK1/␣ and IKK2/␤) (15)(16)(17)(18)(19). IKK␣ and IKK␤ are related molecules of 85 and 87 kDa, respectively, and share 50% sequence similarity. Both proteins contain NH 2 -terminal kinase domains, leucine zipper, and helix-loop-helix motifs (16,19). In vitro phosphorylation studies have shown that both kinases can phosphorylate IB␣ on serines 32 and 36, but IKK␤ is more active in this regard.
Recently, we have cloned, by complementation of an NF-B unresponsive cell line, a third component of the IKK complex (20), that we called NEMO (NF-B Essential Modulator). NEMO is a 48-kDa glutamine-rich protein, which lacks a catalytic subunit, but contains two coiled-coil motifs, a leucine zipper, and a COOH-terminal zinc finger (20). NEMO is a regulatory and structural subunit of the complex which seems to interact directly with IKK␤, but not with IKK␣ (20). Studies with NH 2 -terminal deletion mutants of NEMO show that the first 235 residues contain the site of interaction with IKK␤ (21). The human homolog of NEMO, IKK␥/IKKAP, has been cloned after purification of the IKK complex (21,22).
Data base searching led to the isolation of a cDNA encoding a NEMO-related protein (NRP), which shares 53% sequence similarity with NEMO. NEMO and NRP were also identified in a yeast two-hybrid screen using an adenovirus protein (Ad E3-14.7K) as a bait. Interestingly, these proteins, named FIP-3 and FIP-2, respectively, could block the anti-apoptotic activity of the E3-14.7K protein after TNF␣ stimulation (23,24).
In this study, we have characterized NRP more thoroughly. In order to determine whether NRP is involved in NF-B signaling, we investigated whether it is associated with the IKK complex, or whether it could complement a NEMO-deficient cell line. We could not observe any role of NRP in NF-B signaling, but we found that the COOH-terminal zinc finger of NRP can functionally replace that of NEMO. We could demonstrate that NRP is present in a high molecular weight complex, smaller than the IKK complex, and is associated with two kinase activities. These results demonstrate that NRP could fulfill a similar regulatory function to that of NEMO in a non-NF-B-dependent pathway. Moreover in an effort to characterize the signaling pathways that might target NRP, we * This work was supported in part by grants from the Association pour la Recherche sur le Cancer and the Ligue Nationale Française contre le Cancer (to A. I.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ showed that PMA stimulation induces its phosphorylation. This phosphorylation was inhibited by K-252a and stauroporin. Finally, we also demonstrated that NRP expression is synergistically induced by interferon and TNF␣.

Cells
70Z/3 is a murine pre-B cell line and its variant 1.3E2 is a NEMOdeficient cell line (20). E29.1 is a CD4-negative, CD8-negative mouse T-cell hybridoma (kindly provided by P. Truffa Bachi). Jurkat is a human leukemia T cell line. These cells were maintained in RPMI medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in 5% CO 2 . In the case of 70Z/3 and 1.3E2, 50 M ␤-mercaptoethanol were added in the medium. HeLa cells and 293T cells were maintained in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in 7% CO 2 .

Antisera
The antisera used were the following: anti-IKK2 (Santa Cruz, ref. H470) is a polyclonal rabbit antiserum. Anti-CD28.2 is a murine monoclonal antibody directed against human CD28 (Immunotech). T cells were activated by stimulation with an anti-TCR murine monoclonal antibody (Vit-3) (kindly provided by W. Knapp). Anti-NEMO rabbit polyclonal antibody (serum 44106) was raised against a TrpE fusion protein encompassing amino acids 30 -329 of murine NEMO (20). Anti-NRP (serum 46096) is a polyclonal rabbit antiserum generated against a TrpE fusion protein encompassing amino acids 84 -164 of human NRP.

Plasmids
NRP open reading frames were amplified by the polymerase chain reaction. Expression vectors for transfection into 1.3E2 cells were obtained by subcloning cDNAs encoding NEMO, NRP, or their derivatives into the plasmid pcDNA-3 (Invitrogen). NEMO ⌬ ZF represents amino acids 1-385 of NEMO. For the cloning of NEMO ⌬ ZF-NRP ZF, NRP plasmid was digested with BstBI/EcoRI and this fragment was ligated into the XbaI site of the NEMO plasmid. Details of all the constructions used in this article are available upon request.

Total Extracts of Treated and Non-treated Cells
Treatment with Chemicals-Cells (5 ϫ 10 6 ) were incubated for the indicated time at 37°C in 1 ml of regular growth medium containing 100 IU/ml TNF␣ (Pharmingen), 20 ng/ml IL-1, 10 ng/ml interferon ␣/␤, 10 ng/ml interferon ␥, 15 g/ml LPS (Sigma), 50 ng/ml PMA (Sigma), 1 M calcium ionophore (Sigma), or 50 g/ml cycloheximide (Sigma). Treatment of the cells with kinase inhibitors or activators was performed at 37°C for 20 min and was followed by PMA stimulation for 30 additional minutes. The following kinase inhibitors and activators were purchased from Calbiochem and were used at the indicated doses: K-252a ( Treatment with Anti-TCR and Anti-CD28 -10 ϫ 10 6 Jurkat cells were activated at 37°C with 6 g of the anti-TCR Vit-3 and/or 20 g of the anti-CD28 antibody at 37°C. Lysis Procedure-Cells were lysed by adding an equal volume of 1 ϫ CHRIS buffer (50 mM Tris, pH 8.0, 0.5% Nonidet P-40, 200 mM NaCl, and 0.1 mM EDTA), supplemented with 20 g/ml each of the protease inhibitors leupeptin, aprotinin, and phenylmethylsulfonyl fluoride, as well as the phosphatase inhibitors sodium fluoride (100 mM) and sodium orthovanadate (2 mM).

Secondary Structure Prediction Analysis
This analysis was performed using the ExPASy (Expert Protein Analysis System) proteomics server of the Swiss Institute of Bioinformatics (SIB).

Immunoprecipitations
Cells were lysed as described above. Specific polypeptides were then recovered by immunoprecipitation from equivalent amounts of cellular proteins (1 mg). Immune complexes were collected with Staphylococcus aureus protein A (Pansorbin, Calbiochem). After washing the immuno-precipitates 3 times in lysis buffer, the proteins were resolved by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. Subsequent immunoblots were obtained using the protocol outlined below.

Immunoblots
Immunoblots were performed according to a previously described protocol (25). We used antiserum, biotinylated protein A (Interchim Pierce), and streptavidin-horseradish peroxidase (Amersham Pharmacia Biotech) at 1/1000 dilution. Proteins transferred to Immobilon membranes (Millipore) were revealed with the SuperSignal chemiluminescent substrate (Interchim Pierce) and immunoreactive products were detected by autoradiography.

NF-B Reporter Assays
1.3E2 cells were transfected by the DEAE dextran method. The total DNA content in each transfection was adjusted to 10 g. The Ig-(B) 3 -LUC reporter has been described previously (26). In the control, the cells were transfected with the appropriate empty parental expression vector pcDNA3. 24 h after transfection, LPS was added 5 h before harvest. Then, the cells were lysed on ice for 10 min in 300 l of lysis buffer (25 mM Tris phosphate, pH 7.8, 8 mM MgCl 2 , 1 mM DTT, 1% Triton, 15% glycerol). Cell debris was removed by centrifugation at 13,000 r.p.m. at 4°C for 5 min. The whole cell extract was used to measure luciferase activity in the lysis buffer containing 1 mM Dluciferin (Roche Molecular Biochemicals) and 20 mM ATP.

Immune Complexes Kinase Assay
Cells were lysed with CHRIS buffer as described above. 1 mg of proteins from the lysates were incubated with 5 l of the NEMO or NRP antibody. The immunoprecipitations were performed as described above, except that the immunoprecipitates were washed 3 times with 1 ϫ CHRIS followed by 3 washes with a kinase buffer containing: 10 mM Hepes, pH 7.5, 10 mM MgCl 2 , 100 M Na 3 VO 4 , 20 mM ␤-glycerophosphate, 2 mM DTT, 50 mM NaCl. The kinase reactions were conducted as described (20) for 30 min at 30°C in the kinase buffer, in the presence of 12.5 Ci of [␥-32 P]ATP and GST-IB␣ 1-72 wild type (referred as IB␣ wt) or GST-IB␣ 1-72 A32A36 (IB␣ 2A) as substrates.

RNA Preparation and Northern Blot Analysis
RNA was extracted in TRIzol (Life Technologies), according to the manufacturer's instructions. 10 g of total RNA were separated on 1% denaturing formaldehyde-agarose gel, transferred onto nylon membrane (Amersham Pharmacia Biotech), and hybridized with a 32 Pradiolabeled probe corresponding to full-length human NRP cDNA. Hybridization was carried out at 60°C in 7% SDS, 0.5 M sodium phosphate, pH 7.0. The membrane was washed successively in 0.1% SDS, 2 ϫ SSC at 25°C for 5 min and in 0.1% SDS, 0.1 ϫ SSC at 60°C for 30 min.

Cell Transfection
HeLa cells were trypsinized, pelleted, and resuspended in complete medium at 5 ϫ 10 6 cells/1 ml with a total amount of 20 g of plasmid, and the cells were electroporated with the Easyject system (Eurogentec, Seraing, Belgium) in 4-mm cuvettes. Electroporation parameters were: 240 V, 1350 mivrofarads, 156 ohm, pulse time 211 ms.

Immunofluorescence
Transfected HeLa cells were seeded in 5 ml of minimal essential medium in 6-well plates and grown on glass coverslips. After 24 h the medium was changed. After an additional 24 h the cells were washed in PBS two times, fixed in 3% paraformaldehyde in PBS for 15 min at 37°C, rinsed twice in PBS and permeabilized in phosphate-buffered saline (PBS) containing 0.5% (v/v) Triton X-100 in PBS for 10 min. After two washes with PBS, unspecific hybridization was blocked with 3% (v/v) bovine serum albumin in PBS for 15 min and incubated with primary antibody (purified NRP antibody) at 1/100 for 1 h at 37°C. After washes, coverslips were incubated with Cy3-conjugated secondary antibody (Sigma) diluted 1:200. After being washed in PBS, coverslips were mounted on slides with Mowiol, and cells were examined with a Leica DMRXA-HC microscope.

Preparation of S100 Extracts and Gel Filtration Analysis
Fifty million 293T cells were washed in PBS and resuspended in 500 l of 50 mM Tris, pH 7.5, and 1 mM EGTA. Cells were lysed by 30 passages through a 26-gauge needle. After centrifugation for 10 min at 1,500 rpm, the supernatant was recovered and complemented with 1 mM DTT, 0.025% Brij 35, and a mixture of proteases and phosphatases inhibitors. S100 were prepared by centrifuging the cytoplasmic extracts for 30 min at 52,000 rpm in a TLA 100.2 rotor (Beckman). After adding 10% glycerol, the S100 extracts were quickly frozen in dry ice and stored in liquid nitrogen. Gel filtration chromatography was carried out on a Superose 6 column (Amersham Pharmacia Biotech) precalibrated with aldolase (158 kDa), catalase (232 kDa), ferritin (440 kDa), and thyroglobulin (669 kDa). Five-hundred l fractions were recovered and directly analyzed by Western blotting with anti-NEMO and anti-NRP antisera.

In Gel Kinase Assay
Cells were lysed and proteins were immunoprecipitated as described above. Immunoprecipitates were applied to a 10% polyacrylamide-SDS gel which had been polymerized in the presence of 0.5 mg/ml myelin basic protein (bovine brain, Sigma). Following electrophoresis, the gel was subjected to several rounds of denaturation and renaturation. For the denaturation procedure, the gel was washed twice for 30 min with a 50 mM Hepes, pH 7.6, solution containing 20% isopropyl alcohol, and twice for 30 min with buffer A (50 mM Hepes, pH 7.6, 5 mM ␤-mercaptoethanol), and finally twice with 6 M urea in buffer A. Then the gel was renatured progressively at 4°C by washing the gel once in a 3 M urea solution in buffer A containing 0.025% Tween 20 for 15 min, once in a 1.5 M urea solution in buffer A containing 0.025% Tween 20 for 15 min, and once in a 0.75 M urea solution in buffer A containing 0.025% Tween 20 for another 15 min. For the kinase reaction, the gel was preincubated in kinase buffer (20 mM Hepes, pH 7.6, 20 mM MgCl 2 ) at 30°C for 30 min and finally incubated at 30°C for 2 h in the kinase buffer containing 2 mM DTT, 20 M ATP, and 100 Ci of [␥-32 P]ATP. The gel was then washed 4 times for 15 min with trichloroacetic acid/PP i (5% trichloroacetic acid, 1% NaPP i ), dried, and exposed to x-ray film.

NRP Is Structurally
Homologous to NEMO-Our previous results have shown that NEMO-deficient cell lines are refractory to all tested NF-B-activating stimuli, suggesting that IKK is the unique complex involved in NF-B signaling (20,27). However, the possibility exists that kinase complexes different from the IKK complex might exhibit a tissue-restricted expression, or might respond to uncharacterized NF-B activating stimuli. In an effort to identify components of such complexes, we looked for NEMO homologues by data bank searching. BLAST search allowed us to identify a cDNA encoding a 67-kDa protein we called NRP. NRP was also isolated under the name of FIP-2 in a yeast two-hybrid screen using the adenovirus E3-14.7K protein as a bait (24).
Alignment of the amino acid sequence encoded by NRP with that of NEMO shows 53% amino acid similarity (Fig. 1). NRP consists of 572 amino acid residues with a predicted molecular mass of 67 kDa. Alignment of the 2 sequences indicates that NRP contains an "insert" of 167 additional amino acids positioned after residue 134 of NEMO (Fig. 1). Secondary structure prediction of the 2 proteins shows similar structural elements. Both proteins have coiled coil domains, a carboxyl-terminal zinc finger, and a putative leucine zipper, which is located at the beginning of the insert in NRP and in the COOH-terminal part of NEMO (Fig. 1).
NRP Cannot Reconstitute the NEMO-deficient Cell Line 1.3E2-We have recently reported that NEMO could complement the 70Z/3-derived mutant cell line 1.3E2, that exhibits a defect in NF-B activation (20). As NRP and NEMO are related, we asked whether NRP could also complement the 1.3E2 cell line. Transient transfection of a vector expressing NEMO in 1.3E2 resulted in a 30 -40-fold increase in NF-B activity following LPS stimulation, whereas transfection of the parental expression vector had no effect ( Fig. 2A). In contrast to NEMO, transfection of NRP did not result in NF-B activation under the same conditions, suggesting that NRP could not replace NEMO in NF-B signaling. We recently showed that the zinc finger region of NEMO is essential for its activity, and that its deletion strongly reduces its ability to complement the 1.3E2 cell line ( Fig. 2A). 2 As the zinc finger domains of NRP and NEMO are very similar (Fig. 1), we tested the possibility that the zinc finger domain of NRP could functionally replace that of NEMO. A hybrid molecule consisting of the NRP zinc finger fused to the rest of the NEMO molecule (NEMO ⌬ZF-NRP ZF) could activate NF-B to the same extent as wild-type NEMO, suggesting a possible functional similarity between the two molecules.
NRP Is Not Associated with NEMO or IKK␤-The significant homology between NEMO and NRP led us to investigate whether NRP was associated with the IKK complex. To examine this possibility, E29.1 T cell hybridoma lysates were subjected to immunoprecipitation with normal rabbit serum as a control (Fig. 2B, lane 1), NRP antibody (lane 2), or NEMO antibody (lane 3). As described previously (20), immunoprecipi-2 G. Courtois and S. Yamaoka, manuscript in preparation. tation of NEMO followed by immunoblotting for IKK␤ demonstrated the association between these proteins (lane 3); however, we did not find IKK␤ in NRP immunoprecipitates (lane 2) and we could not detect an association between NEMO and NRP by immunoprecipitating one of these two proteins followed by immunodetection of the other (lanes 2 and 3). Taken together, these results show that NRP is not associated with NEMO, IKK␤, IKK␣ (data not shown) and is probably not a component of the IKK complex.
NRP Is Part of a High Molecular Weight Complex Different from the IKK Complex-Since NEMO is an essential component of the 600 -800-kDa kinase complex that phosphorylates IB, we investigated whether NRP is also present in a high molecular mass complex. S100 extracts were prepared from 293T cells and fractionated on a Superose 6 gel filtration column. The fractions were analyzed by Western blotting using antibodies directed against NRP and NEMO. As shown in Fig. 2C, NRP is present in a high molecular mass complex ranging from 400 to 700 kDa (lanes 7-10), whereas the larger IKK complex containing NEMO migrates, as expected, at 600 to 800 kDa (lanes 9 -11). These results suggest that these two proteins are essentially present in different complexes.
NRP Is Not Associated with an IB␣ Kinase Activity-The results presented above demonstrate that NRP is not associated with known IB kinases but, they do not address whether NRP is associated with other putative IB kinases. In order to elucidate this point, NEMO and NRP were immunoprecipitated from resting or stimulated cells. The immune complexes were assayed for phosphorylation of a glutathione S-transferase (GST) fusion protein containing the amino-terminal part of IB␣ in the wild type context (IB␣) or mutated on its two phosphorylation sites (IB␣ 2A). In Fig. 2D, top panel, Jurkat cells were stimulated for 10 min with PMA and ionomycin (lanes 3 and 4), anti-CD28 (lanes 5 and 6) or both (lanes 7 and 8). As expected, NEMO was found to be associated with an inducible kinase activity as highlighted by IB␣ phosphorylation (compare lanes 3, 5, 7 to lane 1). This activity was directed against the phosphorylation sites of IB␣ since IB␣ 2A was not phosphorylated (lanes 2, 4, 6, and 8). This phosphorylation was maximum when the cells were co-stimulated with PMA, ionomycin, and anti-CD28 (compare lanes 7 to lanes 3 and 5). In contrast, we could not observe any kinase activity associated with NRP (lanes 9 -16). In the bottom panels, we have evaluated the effect of PMA and TNF␣ treatment on NEMO and NRP kinase activity in HeLa and E29.1 cells. We could observe an inducible kinase activity associated with NEMO after TNF␣ stimulation (compare lanes 5 and 1) and to a lesser extent after PMA stimulation (compare lanes 3 and 1); however, we could not observe any kinase activity associated with NRP. These data indicate that NRP, in contrast to NEMO, is not associated with an IB␣ kinase activity.
Effect of Different Stimuli on NRP Expression-In order to investigate the effect of different NF-B stimuli on NRP expression, 70Z/3 preB cells were stimulated for the indicated time with PMA. As depicted in Fig. 3, top panel, precipitation of NRP followed by immunoblotting with the same antibody demonstrated an increase in the appearance of a slower migrating form of the molecule after 15 min of PMA stimulation (lanes [2][3][4][5]. This upper band progressively returned to basal level after 1 h of stimulation (lanes 4 and 5). We then tested whether other stimuli could also increase the amount of the upper band. 70Z/3 cells were subjected to IL-1 (middle panel) or LPS stimulation (bottom panel), two other stimuli able to activate NF-B in this cell line; we could not detect any effect on NRP expression following these treatments. To evaluate whether similar events could occur in T cells, the murine hybridoma E29.1 was subjected to PMA treatment (Fig. 3B, top panel). We could observe a strong increase in the appearance of the slow migrating band after 5 min of stimulation (compare lanes 2 and 1). However, TNF␣ treatment of E29.1 cells did not result in any change in NRP expression (bottom panel), although prolonged treatment with this cytokine strongly increased the level of the NRP mRNA (see below).
PMA Induces Phosphorylation of NRP-We then investigated whether this slowly migrating band represents a hyperphosphorylated form of NRP. E29.1 cells were left untreated or were stimulated for 30 min with PMA and then lysed with Nonidet P-40 containing buffer (Fig. 4, lanes 1 and 2). Following this treatment, cell extracts were treated with -phosphatase (lanes 3 and 4) or with -phosphatase plus phosphatase inhibitors (lanes 5 and 6). PMA treatment induced the appearance of the upper band (lane 2), which disappeared following phosphatase treatment; this disappearance could be blocked by phosphatase inhibitors (lane 4). This clearly demonstrates that the upper bands represent an hyperphosphorylated form of NRP.
Stability of NRP Expression and Phosphorylation-To evaluate the half-life of NRP in stimulated and non-stimulated cells, we used the protein synthesis inhibitor cycloheximide (Fig. 5, top panel). We first treated E29.1 T cells with cycloheximide alone for the indicated period of time (lanes 7-12). This treatment had no effect on the level of expression of NRP until 6.5 h, indicating that NRP has a rather long half-life. In parallel, E29.1 cells were subjected to PMA stimulation (lanes 1-6): this resulted in the appearance of the slowly migrating form described above, and to the slow accumulation of the fast migrating band. The cells were then pretreated with cycloheximide for 30 min and then incubated with cycloheximide and PMA during the indicated times. Inhibition of protein synthesis did not interfere with the appearance of the upper band, but resulted in an accelerated decay of both the slow and fast migrating forms of NRP (compare lanes 13-18 and 1-6). These results suggest that PMA treatment decreases the half-life of NRP.
Since NRP was still phosphorylated after 6 h of PMA stimulation (Fig. 5, lane 6), we then asked whether NRP required an ongoing phosphorylation signal or whether its phosphorylation was stable. To this end, E29.1 T cells were treated as described above with PMA and, after 30 min of treatment (lanes 2 and 7), half of the cells were left with the inducer (lanes 3-5) while the other half were washed twice and resuspended in fresh medium without PMA (lanes 8 -10). Following this "chase" of PMA, the phosphorylated band of NRP only slightly decreased (compare lanes 8 -10 to lanes [3][4][5], suggesting that the phosphorylation of NRP is rather stable even when PMA is removed from the medium. K-252a and Stauroporin Inhibit the Phosphorylation of NRP-In order to characterize the type of kinase leading to NRP phosphorylation, we tested the effect of different kinases inhibitors. As phorbol esters activate protein kinase C (PKC), we first investigated the effect of various PKC inhibitors on the basal or PMA-induced NRP phosphorylation (Fig. 6). Pretreating the cells with GFX (lanes 3 and 4), GÖ 6976 (lanes 5 and 6), Rottlerin (lanes 7 and 8), K-252b (lanes 11 and 12), H7 (lanes 13  and 14), or calphostin C (lanes 15 and 16) showed no effect. Conversely, treatment of the cells with another PKC inhibitor, K-252a, belonging to the family of alkaloid toxin kinase inhibitors, caused a complete inhibition of PMA-induced as well as basal NRP phosphorylation (compare lanes 9 and 10). As K-252a is also a potent inhibitor of CaM kinase II and protein kinase A we evaluated the effect of additional inhibitors or activators of these kinases. Fig. 6, middle panel shows that the calmodulin kinase inhibitors KN-93 (lanes 5 and 6) and KN-62 (lanes 9 and 10) as well as the calmodulin antagonists W7 (lanes 7 and 8) and calmidazolium (lanes 11 and 12) had no effect on PMA-induced NRP phosphorylation, suggesting that CaM kinase II is not involved in NRP phosphorylation. In addition, Bt 2 cAMP, a potent cell-permeable activator of the cAMP-dependent protein kinase (PKA) (lanes 13 and 14) and the pertussis toxin (lanes 15 and 16), which increases intracellular cAMP production increase, were also uneffective in blocking the effect of PMA on NRP. We then evaluated the effect of stauroporin (lower panel, lanes 3 and 4), a potent inhibitor of protein kinases, including CaM kinase, myosin light chain kinase, PKA, PKC, and PKG; genistein (lanes 5 and 6), a protein-tyrosine kinase inhibitor and thapsigargin (lanes 7 and 8), a cell permeable tumor promoting sesquiterpene lactone that releases Ca 2ϩ by inhibiting endoplasmic reticular Ca 2ϩ -ATPase. The cells were also treated with two MAP kinase inhibitors, PD098059 (lanes 9 and 10) and SB203580 (lanes 11 and 12). Fig. 6, lower panel, shows that stauroporin weakly inhibits NRP phosphorylation (compare lanes 4 and 2), genistein has no effect on NRP phosphorylation (lane 6), whereas thapisgargin increases the basal phosphorylation of NRP (compare lanes 7 and 1), suggesting that NRP phosphorylation is Ca 2ϩ -dependent. Finally, it is unlikely that the MAP kinase kinase and p38 play a role in PMA-induced NRP phosphorylation since this event was not modified by PD098059 or SB03580.
In order to determine the sensitivity of PMA-induced NRP phosphorylation to K-252a and stauroporin, we have titrated their effect. We observed that K-252a and stauroporin are efficient in the range of 0.2-2 and 1 M, respectively (data not shown).
NRP Is an Interferon-inducible Protein-As interferon is important for many biological responses, we investigated whether NRP is also inducible by interferon. E29.1 cells were treated with interferon ␥ for different periods of time (Fig. 7A, left  panel). We could not observe any effect on NRP expression at the early time points (lanes 3-8), but interestingly after 18 h of treatment, the amount of NRP was significantly increased (lane 2). The same effect could be observed in HeLa cells (right panel, compare lanes 1 and 2). A similar increase could also be FIG. 3. A, NRP is upshifted following PMA but not IL-1 or LPS treatment in the murine 70Z/3 pre-B cell line. Cells were stimulated for the indicated periods of time with TNF␣, IL-1, or LPS, lysed, and subjected to NRP immunoprecipitation followed by immunoblotting with the NRP antibody. The 2 forms of NRP are indicated on the right by arrows. B, NRP is upshifted following PMA but not TNF␣ treatment in E29.1 T cell hybridoma. Cells were treated with different inducers for the indicated periods of time. After lysis, and immunoprecipitation with NRP antiserum, the proteins were assayed by immunoblotting with anti-NRP antibody. The position of NRP is indicated on the right by an arrow.  3). The transferred Immobilon membrane was cut in 3 parts before immunoblotting. The different antibodies used to blot the corresponding pieces of the membrane are shown on the right. The position of IKK␤, NRP, and NEMO are indicated on the right and the position of the heavy chain of IgG (Ig) is shown on the left. C, gel filtration analysis of NRP and NEMO in 293T cells. S100 extracts were prepared as described under "Materials and Methods" and fractionated through a Superose 6 column. Fractions were subsequently analyzed by Western blotting using antibodies directed against NRP or NEMO. D, NRP is not associated with an IB␣ kinase activity. Jurkat, HeLa, or E29.1 cells were either non-stimulated or stimulated for 10 min with the indicated stimuli, then lysed, and NEMO and NRP proteins were recovered by immunoprecipitation. Kinase activity was assayed by immune complex reactions in the presence of [␥-32 P]ATP and 1 g of GST-IB␣ 1-72 wild type (referred as IB␣ wt) or GST-IB␣ 1-72 A32A36 (IB␣ 2A) as substrates. The position of the phosphorylated substrates are indicated on the right. observed following treatment with interferon ␣/␤ (data not shown). This increase on NRP expression could be blocked by cycloheximide, indicating that it is an indirect effect requiring new protein synthesis (data not shown).
In addition to the effect of interferon on NRP expression, Li and colleagues (24) have found that TNF␣ could also increase the level of NRP mRNA. We therefore tested in HeLa cells whether interferon and TNF␣ could cooperatively increase NRP expression (Fig. 7B) 5 and 6). In order to determine at which level NRP expression is controlled by TNF␣ and interferon, HeLa cells were assayed for expression of NRP by Northern blot analysis (bottom panel). As described by Li and colleagues (24), two NRP mRNAs were detected. In agreement with the Western blot, an increased amount of these transcripts could be observed in HeLa cells following treatment by interferon ␣/␤ (lane 2), interferon ␥ (lane 3), and TNF␣ (lane 4), suggesting that these cytokines regulate NRP expression at the transcriptional level. As shown in lanes 5 and 6, TNF␣ synergizes with interferon to stimulate the transcription of NRP.
NRP Is Associated with the Golgi Apparatus-In order to investigate the subcellular localization of NRP, we performed indirect immunofluorescence experiments using affinity purified antibodies directed against NRP. Li and colleagues (24) have reported that overexpression of NRP in the murine C3HA cell line results in a "bead-like perinuclear structure" of this protein. In our studies we could observe that overexpressed NRP in HeLa cells shows a diffuse distribution all over the cytoplasm with a weak dominant staining in the Golgi apparatus (Fig. 8A). In contrast, we found that endogenous NRP was exclusively associated with the Golgi network (panel B). This result was confirmed by double immunofluorescence with a Golgi marker (data not shown). It is likely that overexpressing NRP resulted in an nonspecific cytosolic localization. We also investigated whether PMA treatment might change the Golgi localization of NRP, as this stimuli triggers the phospho-rylation of this protein (Fig. 3). However, we could not observe any change in the subcellular localization of NRP following this stimulation, suggesting that NRP subcellular localization is not influenced by its phosphorylation state (data not shown).
NRP Is Associated with Two Kinase Activities-We have shown in Fig. 2 that NRP is not associated with IKK␣ or IKK␤. However, if NRP performs a similar function as NEMO in another signaling pathway, one might expect it to associate with kinases. To identify the presence of such kinases, an in gel kinase assay was performed. To this end, E29.1 T cells were activated or not by the addition of PMA for 30 min. Triton X-100 whole cell lysates were subjected to immunoprecipitation using preimmune serum as a negative control (Fig. 9, lanes 1  and 2), NEMO antibody (lanes 3 and 4), or NRP antibody (lanes 5 and 6). The catalytic activity and the size of the putative kinases was then assessed in vitro as described under "Materials and Methods" using an SDS-polyacrylamide gel electrophoresis containing myelin basic protein as an exogenous substrate. No association with kinases could be seen when normal rabbit serum was used for immunoprecipitation (lanes 1 and 2). As expected, the anti-NEMO immunoprecipitate was associated with a kinase activity with a molecular mass of about 85 kDa, which probably corresponds to IKK␣ and IKK␤ (lanes 3  and 4). In the case of NRP, we could detect two bands migrating at approximately 85 and 180 kDa (lanes 5 and 6). As demonstrated above (see Fig. 2), it is unlikely that the 85-kDa associated kinase represents IKK␣ or IKK␤. Thus, the 85-and 180-kDa proteins are two kinases that remain to be identified. In order to investigate if the associated kinase activity is related to the inducible phosphorylation of NRP itself, we performed an in vitro kinase assay using recombinant GST-NRP fusion protein as a substrate. However, we could not detect any PMA-inducible kinase activity directed against GST-NRP, indicating that NRP is not a substrate of its associated kinase activity (data not shown).

DISCUSSION
The IKK complex is required for the inducible phosphorylation of IB proteins on critical serine residues in response to NF-B activating stimuli. Two kinases, IKK␣ and IKK␤, are responsible for the activity of this complex. Recently, we cloned the NEMO protein by genetic complementation of a Tax-transformed rat fibroblast cell line (5R) unresponsive to all tested NF-B activating stimuli (20). This protein interacts with IKK␤ and is a component of the IKK complex. We demonstrated that the kinase subunits IKK␣ and IKK␤ require NEMO in order to become responsive to upstream signals. In the 5R cell line, NEMO is absent and reconstitution of the cells with this protein completely restores NF-B activity. Data base searching with the NEMO protein sequence revealed the presence of human EST clones encoding a NRP. This protein shows significant similarities to NEMO in its primary and predicted secondary structure.
Li et al. (23) have isolated a series of proteins called FIPs that bind to the viral protein Ad E3-14.7K. These proteins include FIP-3 which is identical to NEMO (23) and FIP-2 which is identical to NRP (24). Interestingly, both proteins can reverse the protective effect of E3-14.7K on cell line killing induced by TNF␣ (23,24). As NRP and NEMO share 53% se-quence similarity, it seems likely that these molecules could fulfill a similar function. We first demonstrated that NRP was unable to complement a NEMO-deficient cell line; however, a chimeric molecule made of the NEMO protein with the zinc finger domain substituted with that of NRP could complement this cell line. This suggests the possibility that this domain might be involved in similar functions in the two proteins. We then investigated whether NRP plays a role in NF-B signaling. We first tested whether NRP is part of the IKK complex. An analysis of the endogenous IKK complex revealed that IKK␣, IKK␤, and NEMO co-elute upon gel filtration (20), while NRP elutes in different fractions (Fig. 2C). In co-immunoprecipitation experiments, NEMO was found associated with IKK␤ while we could not find an association of NRP with either of these two proteins (Fig. 2B). Immunoprecipitated NRP was unable to phosphorylate recombinant GST-IB␣ protein after treatment with several NF-B inducers (Fig. 2D). The most striking difference between NEMO and NRP is the presence of an insert in the latter protein. The presence of these additional amino acids led us to speculate that it could prevent NRP interaction with IKK␤. However, in vitro translation of a deletion mutant of NRP lacking the insert gave rise to a protein which was unable to interact with IKK␤ when co-translated. Furthermore, this NRP variant was still unable to complement a NEMO-deficient cell line (data not shown).
Whereas our data strongly suggest that NRP cannot substitute for NEMO and is not an integral component of the kinase complex, it still might be a regulator of NF-B activation. However, transfection of 293T cells with an expression vector for NRP did not result in the transactivation or repression of a B-dependent reporter gene (data not shown).
The possibility remains that NRP could be part of another IKK complex, although NRP does not seem to be associated with an IB kinase activity responsive to TNF, PMA, or anti-CD28. Recently, three groups have identified a new IKK-like kinase, IKKi/⑀ (28 -30). This kinase is 27% identical to IKK␣ and IKK␤ and is part of a high molecular weight complex (28). The IKKi/⑀-containing complex is different from the classical IKK complex (28). As the activity of IKK⑀ is PMA-inducible and this kinase is not associated with NEMO, it seemed worth considering that NRP could play the same role as NEMO in the IKKi/⑀ complex. However, our attempts to co-immunoprecipitate NRP and IKKi/⑀ from extracts of Jurkat and E29.1 cells (stimulated or not with PMA) were unsuccessful, suggesting that these two proteins do not interact, even following stimulation. 3 Another IKK-related kinase, TBK1/NAK, showing a strong homology to IKKi/⑀, has been identified recently (31,32). This protein is 48% identical to IKKi/⑀, and 30% identical to IKK␣ and IKK␤. However, this protein seems to exhibit properties of an IKKK, acting upstream of the IKK complex (31). It was tempting to evaluate the participation of NRP in this complex. As observed with IKKi/⑀, we could not identify any interaction between NRP and TBK1/NAK. 3 While we were testing the effect of different NF-B inducers on NRP expression, we found that NRP was hyperphosphorylated following stimulation by PMA, but not by the other stimuli tested. Some basal phosphorylation could be observed in certain cell lines which was further increased after PMA treatment. We also noticed that NRP exhibits a long half-life which is reduced after PMA treatment suggesting that the stability of NRP might be controlled by this phosphorylation event. NRP is still inducibly phosphorylated in a NEMO-deficient cell line excluding the role of the IKK complex in this event. We tested 3 K. Schwamborn, R. Weil, and A. Israël, unpublished data. Cell lysates were immunoprecipitated using NRP antibody and NRP was detected by Western blotting. The heavy chain of the immunoglobins (Ig) and NRP are indicated on the right. Right panel, HeLa cells were left untreated or treated for 18 h with human interferon-␥. Cell lysates were analyzed by immunoblotting using NRP antiserum. B, NRP expression is increased by TNF␣ and interferon. Top panel, HeLa cells were stimulated for 18 h with different inducers, as indicated, lysed, and subjected to NRP immunoblotting. The position of NRP is indicated on the right by an arrow. Bottom panel, Northern blotting analysis. 10 g of total RNA was used in each conditions. The position of NRP transcripts is indicated on the right. whether PMA treatment results in a modification of the apparent molecular weight of the NRP-containing complex: comparison of fractions from non-induced and PMA-treated 70Z/3 cells demonstrated that the size of the complex was unaffected by the stimulation (data not shown).
To further characterize the kinase activity responsible for NRP phosphorylation, we tested the effect of various kinase inhibitors. Given that phosphorylation of NRP was induced by PMA it seemed natural to test a series of commercially available inhibitors to different PKCs. Although definitive conclusions are difficult to draw from such inhibition studies, they do nonetheless allow us to tentatively exclude some PKCs are major candidates as the NRP kinase. For example, GFX had no effect on the phosphorylation status and as such it seems unlikely that PKC isozymes ␣, ␤ I , ␤ II , ␥, ␦, and ⑀ are involved in PMA-induced NRP phosphorylation. Similarly, neither Gö6976, nor Rottlerin were effective, again consistent with a lack of involvement of PKC␣, ␤ I , ␥, and ␦. In fact, among the PKC and CaM kinase inhibitors tested (Fig. 6) only K-252a and stauroporin were able to block PMA-induced phosphorylation of NRP. Stauroporin inhibits the proteolytically generated catalytic fragment of protein kinase C while having no effect on the binding of phorbol esters to the regulatory domain (33). This is consistent with the report that in vitro K-252a competes with ATP but not with phospholipid or Ca 2ϩ (34). The availability of an apparently specific inhibitor will greatly facilitate future studies on NRP activation. Obviously, one of our goals will be the identification of the PMA-inducible kinase responsible for phosphorylation of NRP.
The present work shows the complexity of the regulation of NRP expression. NRP does not seem to be involved in NF-B signaling, and the intriguing question which remains to be solved is to identify the pathway in which this protein is involved. The strong sequence homology between NRP and NEMO together with the functional similarity of their zinc finger suggest that NRP might have a similar role as NEMO in a different signaling pathway. The additional finding that NRP interacts with two kinases (Fig. 9) indicates that NRP could fulfill, like NEMO, a function in the assembly and activity of these kinases.