RelB cellular regulation and transcriptional activity are regulated by p100.

RelB mediates the constitutive nuclear pool of NF-kappaB transcriptional activity in myeloid and lymphoid cells, which is believed to be secondary to its weak interaction with the classical NF-kappaB inhibitor proteins, the IkappaBs. In other cell types, RelB is located in the cytosol, thus suggesting that RelB is also regulated by an inhibitory protein(s). In this study, it is demonstrated that RelB is associated in the cytosol with p100 but not with IkappaBalpha, IkappaBbeta, IkappaBepsilon, nor p105. Its cytosolic control is not affected by stimuli that lead to RelA nuclear translocation, and RelB nuclear localization is prevented by p100, but not by p105 or IkappaBalpha. Structure function analysis p100-RelB interactions indicates that p100 amino acids 623-900 are required for effective interaction and repression of nuclear translocation and RelB driven NF-kappaB-dependent transcription. Moreover, this carboxyl-portion of p100 contains a nuclear export signal(s), which is required for effective retrieval of RelB from the nucleus. Finally, overexpression of NF-kappaB-inducing kinase, a kinase that has recently been shown to induce p100 processing, possibly through IKKalpha activation, causes nuclear translocation of RelB protein. Thus, these studies indicate that p100 is a bone fide inhibitor of RelB and that this transcription factor may be regulated by NF-kappaB-inducing kinase and/or IKKalpha.

NF-B is a family of ubiquitously expressed transcription factors involved in the regulation of activation, proliferation, differentiation, and death of numerous cell types (1)(2)(3)(4). The NF-B family is composed of proteins that form homodimers or heterodimers with each other, and includes NF-B1 (p50), NF-B2 (p52), RelA, RelB, and c-Rel. The NF-B dimers remain transcriptionally inactive when anchored in the cytosol through association with one of the ankyrin-containing NF-B inhibitors, the IBs: IB␣, IB␤, and IB⑀ (5), or with p100 or p105 (atypical ankyrin-containing inhibitors that share a gene transcript with p52 and p50, respectively) (6 -10). Phosphorylation of the NF-B inhibitor protein at specific serine residues by the IKK complex targets it for ubiquitination and degradation by the proteasome, thus enabling the NF-B dimer to translocate into the nucleus where it can bind to the promoter regions of numerous genes (1,(11)(12)(13)(14)(15). The phosphorylation leading to ubiquitination and degradation of NF-B inhibitor molecules has been well characterized for the typical IB mol-ecules, such as IB␣, -␤, and -⑀ (1,(13)(14)(15)(16)(17). p105 has recently been shown to contain analogous serine sites which are also susceptible to phosphorylation by the IKK complex, and stimuli that target the activation of the IKK complex, such as TNF␣ 1 (18), do lead to the phosphorylation of p105 on these serine residues (19). No such TNF␣-sensitive IKK sites have yet been identified on p100 (19,20).
Recent investigations have given NF-B-inducing kinase (NIK) and IKK␣ (a component of the IKK complex) important roles in the regulation of p100 protein (21,22). When NIK was first characterized as a mitogen activated protein kinase kinase kinase, it was shown to be involved in IKK activation; however, subsequent studies clarified that NIK may not be crucial for IKK activation by inflammatory cytokines (23)(24)(25). Recent work has identified IKK␣, upon activation by NIK, as a specific inducer of p100 processing, causing site-specific phosphorylation, leading to its ubiquitination and subsequent processing to p52 (21,22).
IB inhibitor proteins also exert their control of NF-B by entering the nucleus and retrieving the NF-B heterodimer, thereby escorting it out of the nucleus and into the cytoplasm. This function is mediated by nuclear import sequences located within the ankyrin repeats (26) and by nuclear export signals (NES) (27)(28)(29)(30)(31). Mutation of the NES sequence in IB␣, or inhibition of nuclear export of IB␣ with the pharmacological inhibitor leptomycin B, which targets the nuclear export receptor CRM1, prevents IB␣ from exiting the nuclear compartment of the cell (32)(33)(34). No CRM1-binding export sequences have been identified in the other IB molecules (31).
RelB was originally described as an inhibitor of NF-B activity, as it cannot form homodimers nor directly bind to DNA (35). Yet, when coupled with either p50 or p52, RelB can trigger potent transcriptional activation (36,37). RelB contains an amino-terminal activating domain, not present in RelA or c-Rel, which is required, along with the carboxyl-terminal domain, for effective gene transactivation (38). RelB expression is mainly restricted to lymphoid tissues including the thymic medulla, the periarterial lymphatic sheaths of the spleen, and the deep cortex of the lymph nodes (39). These areas all correspond to regions containing antigen-presenting interdigitating dendritic cells. In fact, in mature B cells, differentiated dendritic cells, and macrophages, RelB has a strong and predominant nuclear localization (39 -43). RelBϪ/Ϫ mice survive in utero, but die as early as 10 days after birth from extensive multiorgan mixed lymphoproliferation, myeloid hyperplasia (excess macrophages and granulocytes), and splenomegaly due to extramedullary hematopoiesis (44). These mice lack mature antigen-presenting dendritic cells and a thymic medulla (45). The studies performed in RelB-deficient mice highlight the fact that RelB is essential to the development and regulation of certain cell types, and that its absence cannot be compensated by the presence of the other NF-B proteins.
RelB regulation differs significantly from the other two NF-B transactivators, RelA and c-Rel. It is not apparently regulated by IB␣ or IB␤ (46). In fact, RelB binds IB␣ only weakly (47), and IB␤ not at all (48). RelB's weak interaction with these IBs, coupled with its constitutive nuclear presence in some cell types, such as the antigen-presenting dendritic cells mentioned above, has led RelB to be classified as the member of the NF-B family that contributes to the constitutive pool of NF-B within a cell (47,49). In mature B cells, RelB is mainly located in the cytosol, and translocates to the nucleus only within several hours following CD40 activation (50 -52) or treatment with platelet-derived growth factors or serum in fibroblasts (53). Because these stimuli also result in an increase in RelB transcription and protein synthesis, it is infered that its nuclear translocation may be secondary to an excess of RelB levels over those of a putative cytosolic inhibitor. What controls RelB's cytosolic retention in certain cell types and not others remains unknown.
In this study we have investigated whether IB molecules, other than IB␣ or IB␤, control the cellular localization of RelB. We report here that p100 is the only IB molecule able to associate with RelB in vivo and inhibit both RelB nuclear localization and RelB-dependent transcriptional activation. Moreover, it is demonstrated that NIK triggers the nuclear translocation of RelB secondary to the degradation of p100.

EXPERIMENTAL PROCEDURES
Plasmids-DNA encoding human RelB or RelA was cloned into the pcDNA3 vector already containing an HA-tag immediately upstream of the transcription start site for the inserted DNA. p100 cDNA encoding amino acids 407-582, 407-623, 407-744, 407-900; amino acids 401-920 of p105; or full-length IB␣ cDNA were cloned into pcDNA3 vector containing a Flag-tag immediately upstream of the transcription start site for the inserted DNA. cDNA encoding amino acids 407-582, 407-623, 407-744, or 407-900 of p100 were cloned into pcDNA3 vector containing a cDNA sequence for green fluorescent protein (GFP) immediately upstream of the transcription start site for the inserted DNA. cDNA encoding human RelB or RelA were cloned into pcDNA3 vector containing sequence for red fluorescent protein (RFP) immediately upstream of the transcription start site for the inserted DNA. The Bluciferase reporter plasmid consists of three NF-B concatamers from the human immunodeficiency virus-long terminal repeat cloned upstream of a concavalin-A minimal promoter driving the expression of luciferase. The TK-renilla-luciferase plasmid contains cDNA encoding Renilla luciferase under the control of an upstream herpes simplex virus-thymidine kinase promoter (Promega, Madison, WI). cDNA encoding p52, RelB, p100, p105, IB␣, IKK␤, or kinase-dead IKK␤ were cloned into pcDNA3 vector. myc-NIK plasmid was obtained from Tularik (South San Francisco, CA). The p105 expression vector was obtained from the National Institutes of Health AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health, from Gary Nabel.
Gene Expression and Cell Extract Preparation-Transfections were performed with FuGENE6 (Roche Molecular Biochemicals, Indianapolis, IN) or LipofectAMINE Plus (Invitrogen) according to the manufacturer's protocol. Cytosolic proteins used in immunoprecipitation experiments were obtained from the supernatant of cells first washed with cold phosphate-buffered saline and then lysed with buffer containing 50 mmol/liter Tris, 150 mmol/liter NaCl, and 0.1% Triton X-100 plus 300 M sodium orthovanadate, 1 mM dithiothreitol, 0.5 mM 3,4-dichloroisocoumarin, 2 M phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 1 g/ml leupeptin, and 1 g/ml pepstatin. Lysates were either subjected to immunoprecipitation, as described below, or to SDS-PAGE followed by electrophoretic transfer to polyvinylidene difluoride membrane and analysis for immunoreactivity with the appropriate antibody. Nuclear and cytosolic proteins were prepared using a modification of the method of Dignam (54). Briefly, cells were washed with cold buffer A (10 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 10 g/ml aprotinin, 2 g/ml leupeptin, 1 g/ml pepstatin, 1 mM dithiothreitol, and 2 M phenylmethylsulfonyl fluoride) and lysed in buffer A plus 0.1% Nonidet P-40 for 10 min on ice. The cytosolic fraction was harvested by centrifuging the lysate at 4500 ϫ g for 3 min and collecting the supernatant. The nuclear pellet was washed twice with buffer A (centrifuging at 4500 ϫ g for 3 min each time). The nuclear proteins were then extracted by resuspending the pellet in 20 mM Hepes, pH 7.9, 0.42 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, dithiothreitol, 3,4-dichloroisocoumarin, and protease inhibitors as in buffer A, plus 25% glycerol for 30 min on ice.
Immunoprecipitation and Western Blotting-100 -300 g of cytosolic protein were incubated with one of the following antibodies: rabbit polyclonal anti-human RelB (raised against amino-terminal peptide PSRRVARPPAAPE), rabbit polyclonal anti-human IB␣ sc-371, rabbit polyclonal anti-human IB␤ sc-945, and rabbit polyclonal anti-human IB⑀ sc-7156 (Santa Cruz, Santa Cruz, CA), rabbit anti-human p100 and rabbit anti-human p105, generated against peptide epitopes within the COOH termini of p100 and p105 (generous gifts from Dr. Nancy Rice) (10), mouse monoclonal anti-HA, clone 12CA5 (Roche Molecular Biochemicals), or anti-Flag (Sigma) for 1-2 h at 4°C, after which protein A-agarose beads (Invitrogen) were added for 1 h at 4°C. The beads were then washed three times with lysis buffer and heat denatured to separate proteins from beads. Immunoprecipitated proteins were subject to SDS-PAGE and electrophoretically transferred to polyvinylidene difluoride membrane. Immunoblotting was performed with specific antibodies and visualized by using the ECL Western blotting detection kit (Amersham Bioscience, Inc., Buckinghamshire, UK). Antibodies used for immunoblotting include those listed above in addition to rabbit polyclonal anti-RelB sc-226, rabbit polyclonal anti-RelA sc-109, rabbit polyclonal anti-p50 sc-7178, rabbit polyclonal anti-p52 sc-298 (Santa Cruz, Santa Cruz, CA), and mouse monoclonal anti-histone (Qiagen, Valencia, CA).
Immunofluorescence-HeLa cells were plated onto washed glass coverslips in six-well culture plate chambers. 24 h later they were transfected by FuGENE6 (Roche Molecular Biochemicals) according to the manufacturer's protocol. Immunofluorescence was performed 18 h after transfection as follows. Cells were washed 3 times with phosphatebuffered saline at 37°C, fixed in phosphate-buffered saline plus 2% paraformaldehyde for 10 min at 37°C, washed again, and quenched in phosphate-buffered saline containing 50 mM glycine for 5 min at room temperature. Cells were washed again, permeabilized with methanol for 2 min at room temperature, washed, and blocked with phosphatebuffered saline containing 5% goat serum and 5% glycerol for 30 min at room temperature. Blocking was done only for those experiments using anti-HA antibody. After blocking, cells were again washed and incubated with anti-HA antibody (2 g/ml, Roche Molecular Biochemicals) in phosphate-buffered saline containing 0.09% sodium azide, 1% bovine serum albumin, 0.1% Tween 20, and 10 g/ml DAPI (4Ј,6-diamidino-2phenylindole) (Sigma) for 1-2 h at room temperature, followed by washing and further incubation with a fluorescein isothiocyanate-conjugated goat antibody to rat IgG (ICN, Costa Mesa, CA). Incubation mixtures for experiments visualizing GFP or RFP constructs did not include the anti-HA antibody. Following antibody and/or DAPI incubation, cells were washed, inverted, and mounted onto microscope slides. Slides were viewed by confocal microscopy at ϫ400 magnification with UV, HeNe, and argon lasers for excitation.
Inhibition of Nuclear Export-Cells studied for inhibition of nuclear export were incubated with 200 ng/ml leptomycin B (generous gift from Dr. Minoru Yoshida, Tokyo, Japan) for 6 h before preparation for immunofluorescence.
Gene Transfection and Reporter Assays-HeLa cells were plated 24 h before transfection at 50% confluency into 24-well plates, and then transiently transfected by FuGENE6 (Roche Molecular Biochemicals) according to the manufacturer's protocol. Transfected HeLa cells were lysed in Passive Lysis Buffer supplied in the Dual-Luciferase TM Reporter Assay System (Promega) according to instructions in the accompanying technical manual. Briefly, 10-l lysates were mixed with 100 l of luciferase Assay Reagent II (Promega) and luminescence was measured with a Berthold Lumat to analyze firefly luciferase levels. Then 100 l of Stop & Glo TM Reagent (Promega) was added and luminescence was again measured to analyze Renilla luciferase levels. Relative luciferase units are equivalent to firefly luciferase normalized to Renilla luciferase. All transfection experiments were performed in duplicate.

RESULTS
RelB Physically Associates with p100 -To determine the ability of RelB to physically interact with previously identified NF-B inhibitors, RelB and the various IB molecules were immunoprecipitated from HeLa cell lysates, and the immunoprecipitates were analyzed by immunoblot using either anti-RelB antibodies or antibodies recognizing the corresponding IB molecules. In the case of p100 and p105, these molecules were immunoprecipitated using antibodies directed against the COOH terminus of each molecule, to avoid immunoprecipitating p52 or p50, respectively. As shown in Fig. 1A, RelB is present in the p100 immunoprecipitate (lane 3) but not in the immunoprecipitates of the other IB molecules. Moreover, p100 is present in the RelB precipitate (Fig. 1A, lane 1), further stressing the native and specific association between RelB and p100. RelB does not physically associate with p105, IB␣, IB␤, and IB⑀ (Fig. 1A, lanes 2 and 4 -6). Similar results were obtained in other cell types, including the monocytic cell line U937 and the B lymphocyte cell line BJAB (data not shown).
TNF␣ Targets IB␣ and p105, but Not p100, and Results in the Nuclear Translocation of RelA, but Not RelB-Having demonstrated that p100 is physically associated with RelB, we next investigated whether stimuli, such as TNF-␣, that induce the nuclear translocation of RelA by triggering the degradation of IB molecules like IB␣, result in the nuclear translocation of RelB. HeLa cells were treated with TNF-␣ for up to 100 min, followed by the isolation of nuclear extracts from the cytosolic compartment. Nuclear and cytosolic extracts were analyzed in SDS-PAGE and subjected to immunoblot using antibodies against RelB, RelA, IB␣, p100, and p105. While TNF-␣ triggers a rapid nuclear translocation of RelA, it does not trigger the nuclear translocation of RelB (Fig. 1B). Previously identified degradation of both IB␣ and p105 are observed to temporarily coincide with RelA nuclear translocation, while p100 levels and migration remain unchanged. Similar results were obtained in the promonocytic cell line U937 (data not shown). This implies that TNF-␣Ϫmediated activity on the IKK complex, known to trigger IB␣ and p105 degradation, does not target p100 degradation; this is in direct correlation with absent nuclear translocation of RelB. Altogether, this data further establishes the connection between p100/RelB association in vivo and lack of the TNF-␣-induced nuclear translocation of RelB in the absence of p100 modification and degradation.
p100 Associates with RelB-The significant homology between p100 and p105 suggests that RelB may also interact with p105 in a similar manner as that observed with p100. However, results shown in Fig. 1A indicate that native RelB is constitutively associated in the cytosol with p100, and not p105. To further confirm this, we investigated the interaction between ectopic expression of tagged RelB with the inhibitor proteins FIG. 1. RelB is associated with p100, and TNF-␣ does not target p100 for degradation. A, association between RelB and p100. 100 g of HeLa cellular lysates were immunoprecipitated with antibodies reactive against RelB, a carboxyl-terminal epitope of p105, a carboxyl-terminal epitope of p100, IB␣, IB␤, or IB⑀. Immunoprecipitates were subjected to SDS-PAGE and membranes were blotted with antibodies reactive against RelB, p105, p100, IB␣, IB␤, and IB⑀. B, TNF-␣ stimulation causes degradation of IB␣ and p105, but not p100, and nuclear translocation of RelA, but not RelB. HeLa cells were treated with 10 ng/ml TNF-␣ for various time points. Nuclear and cytosolic compartments were isolated and 20 g of each lysate were subjected to SDS-PAGE. Membranes were blotted with antibodies reactive against RelB, RelA, IB␣, p100, and p105. p100 Inhibits RelB p100⌬N and p105⌬N. p100⌬N is the carboxyl-terminal portion of p100, containing its entire inhibitory ankyrin region, amino acids 407-900; p105⌬N is the carboxyl-terminal of p105, containing its entire inhibitory ankyrin region, amino acids 401-920. We chose to examine the interaction of RelB with only the inhibitory ankyrin containing regions of p100 and p105, to dissect it from the potential interaction of RelB with the Rel homology domain present in the amino-terminal portions of these atypical inhibitor proteins, i.e. p52 and p50, respectively. Co-expression of HA-RelB with Flag-p100⌬N, followed by immunoprecipitation of RelB with anti-HA antibodies and immunoblotting with anti-Flag antibodies, demonstrates that p100⌬N is associated with RelB ( Fig. 2A, lane 5). However, co-expression of HA-RelB with Flag-p105⌬N, followed by HA-RelB immunoprecipitation, indicates that RelB does not associate with p105⌬N ( Fig. 2A, lane 4). When the same experiment was performed expressing HA-RelA, instead of HA-RelB, it is demonstrated that HA-RelA can associate with either Flag-p100⌬N or Flag-p105⌬N ( Fig. 2A, lanes 2 and 1, respectively). This indicates that the lack of association between p105⌬N and RelB is not due to an abnormal configuration of the expressed p105 transgene, as it is shown that p105⌬N effectively interacts with RelA.
The observation that RelA, but not RelB, binds with equal affinity to p105⌬N or p100⌬N was surprising and prompted us to investigate whether, in the presence of both p105⌬N and p100⌬N, RelA or RelB would preferentially interact with one of the IB molecules over the other. This was achieved by coexpressing both Flag-p100⌬N and Flag-p105⌬N together with either HA-RelA or HA-RelB. Cells were then lysed, and cytosolic extracts were subjected to immunoprecipitation by anti-HA antibodies, followed by immunoblotting with anti-Flag antibodies, to determine which of the two IB molecules, p100⌬N or p105⌬N, preferentially interact with either RelA or RelB. Results from these experiments indicate that RelA preferentially associates with p105⌬N ( Fig. 2A, lane 3), while RelB preferentially associates with p100⌬N ( Fig. 2A, lane 6) when both inhibitor molecules are expressed at equivalent levels ( Fig. 2A, lower panel). Surprisingly, RelB levels in the cytosol are significantly decreased in the presence of co-expressed p105⌬N, but not p100⌬N ( Fig. 2A, lower panel, lane 4 versus 5 or 6), suggesting that the inability of p105⌬N to maintain RelB in the cytosol may allow the overexpressed RelB to localize in the nucleus when it is co-expressed with p105⌬N. This potential explanation is supported by results discussed in Fig. 4. Low cytosolic RelB levels prevent an absolute comparison of RelB affinity to p100⌬N or p105⌬N. However, the competitive coimmunoprecipitation indicates a preferential association of RelB with p100⌬N, (Fig. 2A, lane 6). Furthermore, functional experiments to follow (Fig. 4) support the increased affinity of RelB for p100⌬N over p105⌬N.
We next investigated the ability of the classical IB molecule, IB␣, to associate with RelB, for previous data have indicated its weak association with RelB (47). Moreover, coimmunoprecipitation data from Fig. 1A demonstrated that native RelB does not interact with native IB␣, nor any of the other IB molecules, IB␤ and IB⑀. Co-expression of HA-RelB with Flag-p100⌬N, followed by immunoprecipitation of RelB with anti-HA antibodies and immunoblotting with anti-Flag antibodies, demonstrates once more that p100⌬N is associated with RelB (Fig. 2B, lane 5). However, co-expression of HA-RelB with Flag-IB␣, followed by HA-RelB immunoprecipitation and anti-Flag immunoblot, indicates that RelB does not associate with IB␣ (Fig. 2B, lane 4). When the same experiment was performed expressing HA-RelA, instead of HA-RelB, it is demonstrated that HA-RelA can equally associate with Flag-p100⌬N and Flag-IB␣ (Fig. 2B, lanes 2 and 1, respectively), similarly to what was observed for p105. Again, this indicates that the lack of association between IB␣ and RelB is not due to an abnormal configuration of the expressed IB␣ transgene, for we show that IB␣ effectively interacts with RelA. This observation was further investigated by co-expressing, within the same cell, both Flag-p100⌬N and Flag-IB␣ together with either HA-RelA or HA-RelB. Cells were then lysed, and cytosolic extracts were subjected to immunoprecipitation by anti-HA antibodies to determine which of the two IB molecules, p100⌬N or IB␣, preferentially interact with either RelA or RelB. Results from these experiments indicate that RelA preferentially associates with IB␣ (Fig. 2B, lane 3), while RelB preferentially associates with p100⌬N (Fig. 2B, lane 6) when both inhibitor molecules are present at equivalent levels (Fig.  2B, lower panel). Again, as seen when co-expressed with p105⌬N ( Fig. 2A, lane 4), RelB levels in the cytosol are significantly decreased in the presence of co-expressed IB␣, but not of p100⌬N (Fig. 2B, lower panel, lane 4 versus 5 or 6). This could be explained by the inability of IB␣ to maintain RelB in the cytosol, allowing it to localize in the nucleus. This hypothesis is supported by the next set of results summarized in Fig.  4. Thus, RelA preferentially interacts with IB␣ (Fig. 2B, lane  3), while RelB preferentially interacts with p100⌬N (Fig. 2B,  lane 6).
RelB does not form homodimers, and only heterodimerizes with either p50 or p52 (36,37,55). Since the true interaction between the RelB heterodimer and the ankyrin-containing portion of the inhibitor molecule may require the presence of a dimer partner, or at least the amino-terminal portion of the inhibitor, we wished to confirm the results we obtained in Fig.  2 by repeating the competitive co-immunoprecipitations in the presence of overexpressed p52 dimer partner. p52 was chosen as dimer partner in this experiment, for it is the amino-terminal portion of the p100 protein, which has thus far been shown to preferentially associate with RelB. Nevertheless, similar experiments were also performed using p50. HeLa cells were transfected with p52, either HA-RelA or HA-RelB, and either both Flag-p100⌬N and Flag-IB␣ together, or both Flag-p100⌬N and Flag-p105⌬N together. Cells were then lysed, and cytosolic extracts were subjected to immunoprecipitation by anti-HA antibodies to determine which of the IB molecules preferentially interact with either RelA or RelB when p52 is co-expressed. Results from these experiments confirm the previous findings, and indicate that RelA preferentially associates with IB␣ or p105⌬N, while RelB preferentially associates with p100⌬N (data not shown).
We next addressed the ability of full-length tagged p100 or p105 to bind to RelA or RelB. Co-expression of HA-tagged RelA or HA-tagged RelB with Flag-tagged p100 and Flag-tagged p105, followed by immunoprecipitation with anti-HA antibodies and immunoblotting with anti-Flag antibodies, demonstrates that both RelA and RelB can associate with full-length p100 or full-length p105 (Fig. 3A). Because of the difficulty in separating full-length p100 from full-length p105 by gel electrophoresis, additional immunoprecipitations were blotted with antibodies specific for the N terminus of p100 (Fig. 3A, lanes 5 and 6) or p105 (Fig. 3A, lanes 7 and 8). Results from these experiments indicate that RelB preferentially associates with p100 (Fig. 3A, lanes 5 and 6) over p105 (Fig. 3A, lanes 7  and 8), similar to that observed with the ⌬N p100 truncated protein.
Co-expression of HA-RelB with Flag-p100 and Flag-IB␣, followed by immunoprecipitation of RelB with anti-HA antibodies and immunoblotting with anti-Flag antibodies, demonstrates that RelB preferentially associates with full-length p100 Inhibits RelB p100 over IB␣ (Fig. 3B, lane 4). However, when the same experiment was performed expressing HA-RelA, instead of HA-RelB, it is demonstrated that HA-RelA associates with either Flag-IB␣ or Flag-p100 (Fig. 3B, lane 3). These results demon-strate that in an overexpression system, RelA will associate with IB␣, p100, or p105 while RelB will associate preferentially with p100 or p105. The association of RelB to full-length p105 is likely due to affinity for the Rel homology domain in the p100 Inhibits RelB NH 2 terminus since no association of RelB was found with the COOH terminus of p105 ( Fig. 2A, lanes 4 and 6). Altogether, these data confirm the findings with endogenous proteins (Fig.  1A) indicating that RelB preferentially associates with p100 over the other IB proteins.
p100, but Not p105 nor IB␣, Inhibits RelB Nuclear Localization-Based on the above, we hypothesized that p100 is a bone fide cytosolic inhibitor of RelB as compared with p105 or IB␣. To confirm this, we analyzed the ability of RelB to localize in the nuclear compartment when overexpressed in the presence of p100, p105, or IB␣. HeLa cells were co-transfected with HA-RelB or HA-RelA and either full-length p100, p105, or IB␣ cDNA. Transfected cells were fixed and incubated with anti-HA antibodies followed by a fluorescein-conjugated secondary antibody, stained with DAPI to identify the nuclear compartment, and analyzed by confocal microscopy. Only a fraction of the cells are transfected, yet all cells stain blue for DAPI. As shown in Fig. 4, HA-RelB is found in the nuclear compartment when expressed in the absence of other inhibitor molecules (box 1). Co-expression of p100, but not p105 or IB␣, blocks RelB presence in the nucleus (Fig. 4, boxes 2, 3, and 4,  respectively). Rather, co-expression with p105 or IB␣ allows HA-RelB to undergo a nuclear localization. HA-RelA likewise localizes to the nucleus when expressed alone (Fig. 4, box 5), yet it is retained in the cytosol when co-expressed with p100, p105, or IB␣ (Fig. 4, boxes 6, 7, and 8, respectively). Thus, while RelA nuclear localization is inhibited by p100, p105, or IB␣ RelB nuclear presence is only inhibited by p100. The inability of p105 and IB␣ to retain RelB in the cytosolic compartment (Fig. 4, boxes 3 and 4) as seen by immunofluorescence, confirms that the low expression of HA-RelB in the cytosol seen in Fig.  2 is indeed due to the inability of both p105⌬N and IB␣ to prevent the nuclear localization of HA-RelB. This experiment was also performed using p100⌬N or p105⌬N instead of fulllength protein, and similar results were observed (data not shown). Altogether, this experiment indicates that RelB nuclear localization is preferentially inhibited by p100.
The p100 Carboxyl-terminal Domains Are Required for p100 Interaction with RelB and Inhibition of RelB-dependent Transcriptional Activity-To identify which domain(s) of p100 is/are responsible for its association with RelB, a series of 3Ј to 5Ј deletions were introduced into either the Flag-p100⌬N or GFP-p100⌬N (Fig. 5A). The various Flag-tagged expression vectors were then co-transfected with HA-RelB into HeLa cells. Cells were lysed and cytosolic extracts were immunoprecipitated for HA-RelB using anti-HA antibodies. The immunoprecipitates FIG. 3. RelB associates with fulllength p100 and p105 but not IKB␣. 293 T cells were transfected with expression vectors coding for HA-RelA, HA-RelB, Flag-IB␣, Flag-p100, and Flag-p105 as indicated. Anti-HA immunoprecipitations were performed and run on polyacrylamide gels along with 10 g of cytosolic extracts. The proteins were transferred to polyvinylidene difluoride membranes and blotted with antibodies specific for the HA or Flag tags (A and B) as well as antibodies to the NH 2 terminus of p100 or p105 (A).
The functional relevance of the inhibition of RelB nuclear localization by the p100⌬N construct 407-900 was further an-alyzed by measuring the transcriptional activity of RelB in the presence or absence of the p100⌬N constructs. HeLa cells were co-transfected with an NF-B concatemer driving firefly luciferase, a minimal TK promoter driving renilla luciferase, and RelB and p52 expression vectors in the presence or absence of the different p100⌬N deletion constructs (407-582, 407-623, 407-744, or 407-900). Cells were lysed and measured for normalized firefly luciferase activity. As shown in Fig. 6A, p100⌬N constructs 407-900, and to a lesser degree 407-744, inhibited the transcriptional activity driven by RelB/p52 in a dose-dependent manner. However, p100⌬N constructs 407-582 and 407-623 were unable to prevent RelB/p52-dependent transcriptional activity. Similar results were seen with RelB expressed without its dimer partner p52 (Fig. 7B). Moreover, full-length p100 was also seen to inhibit the transcriptional activity driven by RelB or RelB/p52 in a dose-dependent manner (Fig. 7C).
Cytosolic Localization of RelB by p100⌬N Is in Part Dependent on Its Nuclear Export-To explain the inability of p100⌬N construct 407-623, and to a lesser extent 407-744, to inhibit neither RelB nuclear translocation (Fig. 6) nor RelB transcriptional activity (Fig. 7), despite being able to physically associate with RelB (Fig. 5B), we next studied the ability of the various p100⌬N constructs to effectively shuttle between the nuclear and cytosolic compartments. To do so, we first determined the cellular location of each GFP-p100⌬N protein. HeLa cells were transfected with GFP-p100⌬N constructs containing amino ac- FIG. 4. RelB nuclear localization is inhibited by p100, but not by p105 or IB␣. Top panels, HeLa cells were transfected with 500 ng of HA-RelB cDNA, either alone (box 1), with 500 ng of p100 cDNA (box 2), with 500 ng of p105 cDNA (box 3), or with 500 ng of IB␣ cDNA (box 4). Bottom panels, HeLa cells were transfected with 500 ng of HA-RelA cDNA, either alone (box 5), with 500 ng of p100 (box 6), with 500 ng of p105 cDNA (box 7), or with 500 ng of IB␣ cDNA (box 8). Cells were fixed and probed with anti-HA antibodies followed by fluorescein-conjugated secondary antibodies. Nuclei are stained with DAPI.
p100 Inhibits RelB ids 407-623, 407-744, or 407-900. Cells were then fixed and visualized by confocal microscopy for the subcellular localization of the GFP-p100⌬N deletion constructs. Deletion constructs truncating at amino acid 623 or 744 were located within the nuclear compartment of HeLa cells (Fig. 8, rows 1 and 2), but a construct containing the entire carboxyl-terminal, ankyrin-containing region, i.e. amino acids 407-900, was located entirely in the cytosolic compartment of the cell (Fig. 8,  row 3).
Based on the above result, we asked whether the cytosolic localization of p100⌬N 407-900 was due to the presence of putative NES that are not present in p100⌬N constructs truncated prior to amino acid 744. For this, HeLa cells transfected with GFP-p100⌬N 407-900 were treated or not with a pharmacological inhibitor of the CRM1-mediated nuclear export, leptomycin B. Treatment with LMB resulted in GFP-p100⌬N 407-900 being localized to the nuclear compartment (Fig. 8, row 4, as compared with Fig. 8, row 3). Moreover, HeLa cells transfected with GFP-p100⌬N 407-623 or GFP-p100⌬N 407-744, and treated with leptomycin B, resulted in no discernable change in the nuclear presence of these constructs (data not shown). Thus, the cytosolic localization of p100⌬N, and therefore, the ability of p100 to inhibit RelB transcriptional activity, appears to rely upon the ability of p100 to shuttle between the nuclear and cytoplasmic compartments of the cell secondary to one or more NES located between amino acids 744 and 900 of the p100 protein.
NIK Causes p100 Degradation and RelB Nuclear Localization-To confirm the functional relevance of p100 control over RelB subcellular localization, we reasoned that inducing the degradation of p100 should result in the nuclear translocation of RelB. For this, we exploited the recent finding that p100 degradation can be induced by NIK (21). HeLa cells were cotransfected with either control vector, RelB alone, or RelB plus p100, alone or in the presence of NIK, IKK␤, or kinase-dead IKK␤. Nuclear and cytosolic compartments were separated, analyzed by SDS-PAGE, and subjected to immunoblotting using antibodies against RelB, p100, histone, and ␤-actin (Fig.  9A). HeLa cells transfected with RelB alone had detectable RelB in the nuclear compartment (Fig. 9A, lane 2), which was decreased when p100 was co-expressed (Fig. 9A, lane 3). However, co-expression of RelB and p100 with NIK resulted in significant degradation of p100, and this was accompanied by movement of RelB to the nucleus (Fig. 9A, lane 4). Co-expression of RelB, p100, and IKK␤ demonstrates that IKK␤ has no effect on p100 processing or RelB nuclear localization (Fig. 9A,  lane 6). Moreover, when kinase-dead IKK␤ is co-transfected with NIK, p100, and RelB, this mutant kinase does not interfere with NIK-induced p100 degradation and RelB nuclear translocation (Fig. 9A, lane 5), confirming the lack of involvement of IKK␤ in targeting p100 for degradation. These results correlate with and provide an explanation for the inability of TNF␣ to either degrade p100 or translocate RelB into the nucleus, as seen in Fig. 1B, for TNF␣ is targeting IKK␤ activity and, apparently, not NIK. In addition, this set of results confirms that full-length p100 is an effective inhibitor of RelB and that it is susceptible to regulation by NIK. Expression of IKK␤ does not result in lower p100 levels in the cytosol, nor in RelB nuclear translocation (Fig. 9A, lane 6).
To extend the findings shown in Fig. 8A, we decided to confirm the influence of NIK on p100 degradation and RelB cellular localization using immunofluorescence. HeLa cells were transfected with the RFP-tagged RelB, its dimer partner p52, and green fluorescent protein-tagged p100 deletion mutant 407-900 with or without NIK. Transfected cells were then subjected to immunofluorescence studies. Only a fraction of cells were transfected with the RFP and GFP constructs, yet all nuclei stained blue with DAPI. As shown in Fig. 9B, and as seen before, RFP-RelB co-expressed with GFP-p100⌬N 407-900 (row 1) was retained in the cytoplasm (Fig. 9B, row 1), p100 Inhibits RelB whereas the addition of NIK prevented RFP-RelB's cytosolic retention and allowed RFP-RelB to localize in the nucleus (row 2). Furthermore, in these panels, and in others not shown here, the addition of NIK caused a decrease in cytosolic levels of GFP-p100⌬N, implying that GFP-p100⌬N is being degraded in a manner that is dependent on NIK activity.    (56). It is therefore important to dissect potential differences between how RelB and other NF-B family members, RelA or c-Rel, are regulated. In this study, we have demonstrated that RelB cellular localization is regulated by p100 and not by other IB proteins. We have further demonstrated that RelB is not regulated by IKK␤ but by NIK or NIK activation of IKK␣ through its targeting of p100. Such findings strongly argue for a difference in regulation of RelA and RelB. Furthermore, we have identified the regions of p100 required for effective association with and retention of RelB in the cytosol, and determined that p100 Inhibits RelB p100 contains one or more nuclear export signals which are crucial for the cytosolic localization of p100, and thus, functional inhibition of RelB.
Previous studies had demonstrated that IB␣ binds weakly to RelB as compared with RelA (47), that the COOH terminus of p100 can inhibit transcriptional activity mediated by RelB (57), and that p100 retains RelB in the cytosol of specific breast cancer cell lines (58). Results from the present study extend those observations by formally demonstrating the differences in affinity of the different IB molecules toward RelA and RelB, and further identifying molecular mechanisms whereby p100 inhibits RelB nuclear localization.
Until now, p100 and p105 have always been considered to similarly function as alternative inhibitors of NF-B, but this study highlights significant functional differences between them. First, while significant homology exists between these . Nuclear and cytosolic compartments were isolated and 10 g of each lysate were subjected to SDS-PAGE. Membranes were blotted with antibodies reactive against RelB, histone, p100, and ␤-actin. B, HeLa cells plated onto coverslips were transfected with 3 g of RFP-RelB cDNA, 500 ng of p52 cDNA, and 500 ng of GFP-tagged p100⌬N deletion constructs containing amino acids 407-900 cDNA with (row 2) or without (row 1) 500 ng of Myc-NIK. Nuclei are stained with DAPI. Cellular location of RFP-and GFP-tagged proteins was determined by immunofluorescence. Column 1, location of RFP-RelB; column 2, location of GFP-p100⌬N deletion constructs; column 3, nuclei stained with DAPI; column 4, overlay of RFP, GFP, and DAPI. p100 Inhibits RelB two IB-like molecules, RelB has a significant affinity toward p100 and much less toward p105, yet both inhibitors can bind RelA. This suggests that RelB may seek a unique protein conformation in p100, most likely conferred by ankyrin region 4, that favors its interaction with p100. Second, recent data indicate that TNF␣ stimulation of the IKK complex can lead to the inducible phosphorylation of not only IB␣ and IB␤ but also p105, but not p100 (19). It appears p100 phosphorylation and degradation is dependent upon specific IKK␣ activity, possibly induced via NIK (19,21,22). This data is further highlighted by the inability of TNF-␣, which induces IKK activity and leads to the translocation of RelA to the nucleus, to result in the nuclear translocation of RelB. This corresponds to the finding that TNF␣ stimulation leads to p105 degradation, but does not affect p100 (59).
The demonstration that p100 prevents the nuclear localization and transcriptional activity of RelB, dependent upon the nuclear export of p100 is, to our knowledge, the first report of an atypical inhibitor of NF-B containing a nuclear export signal. The precise identification, and mutational analysis, of this NES will undoubtedly be the next step in describing this mechanism of p100 nuclear and cytosolic shuttling. Efforts are underway to identify this sequence, and have been thus far delayed only by the identification of more than one potential NES in the carboxyl-terminal portion of p100 between amino acids 744 and 900. It is possible that a series of NES located in this portion of p100 act in concert to facilitate nuclear shuttling. To our knowledge, the identification of a sequence acting as a nuclear localization signal has yet to be identified on the p100 protein, and may prove an interesting area of investigation.
There is an ongoing debate regarding how p52 and p100 are generated from the same RNA transcript. The established mechanism is that p100 protein is translated, and p52 is simply a product of p100 protein processing (20). This mechanism offers the possibility that the ankyrin-containing carboxyl-terminal region of p100, i.e. p100⌬N or IB␦, may be generated from this protein processing. A second mechanism has recently been offered which instead describes p100 and p52 as being alternatively generated from the same transcript in a co-translational manner, dependent upon proteolytic processing by the proteasome and a glycine-rich region in the carboxyl-terminal portion of p52 (60). Regardless of the functional mechanism, in this study we have demonstrated that both full-length p100 and its COOH terminus are effective in inhibiting RelB nuclear localization. We believe that the experimental results from the coimmunoprecipitation of native p100 with RelB (Fig. 1A), the confocal microscopy results in Fig. 4, and the ability of fulllength p100 to inhibit the transcriptional activity of RelB or RelB/p52 on an NF-B-driven luciferase reporter (Fig. 7C) is compelling evidence that the full-length p100 protein is indeed the physiological inhibitor of RelB. Thus, the p100⌬N constructs we have created may be useful reagents to inhibit RelB in cell and disease models in which RelB protein is predominantly present in the nucleus, such as in B cell malignancies including Epstein-Barr virus-infected B cells or in other antigen presenting cells, such as macrophages and dendritic cells.
Finally, the finding that RelB cellular localization is controlled by p100 lends potential insight into a set of lymphomas that correlate with a translocation or deletion of the NFKB2 gene, which codes for p100. The mutations that occur within this class of lymphomas lead to a truncation of the p100 protein that can no longer repress certain NF-B activity (61-64). We hypothesize that this unrepressed NF-B activity may correspond to unregulated RelB activity, and that p100s lack of repression on RelB activity may be what drives the lymphoproliferation seen in these diseases. This is initially supported by the finding that the shorter 3Ј p100 deletions tested in the present study are less able to inhibit RelB nuclear translocation or its transcriptional activity by virtue of their inability to exit the nucleus of the cell. If this hypothesis holds true and these lymphomas are found to have elevated RelB activity, then the finding that p100 is RelB's physiological inhibitor will have clinical relevance in certain lymphomas, and future treatments for these lymphomas could be appropriately directed to controlling RelB nuclear presence and transactivation.