Negative Regulation of Transactivation Function but Not DNA Binding of NF-κB and AP-1 by IκBβ1 in Breast Cancer Cells*

The transcription factor NF-κB regulates the expression of genes involved in cancer cell invasion, metastasis, angiogenesis, and resistance to chemotherapy. In normal cells NF-κB is maintained in the cytoplasm by protein-protein interaction with inhibitor IκBs. In contrast, in cancer cells a substantial amount of NF-κB is in the nucleus and constitutively activates target genes. To understand the mechanisms of constitutive NF-κB activation, we have analyzed the function of IκBα and IκBβ in breast cancer cells. In most cases, constitutive NF-κB DNA binding correlated with reduced levels of either IκBα or IκBβ isoforms. Overexpression of IκBα but not IκBβ1 resulted in reduced constitutive DNA binding of NF-κB in MDA-MB-231 cells. Unexpectedly, IκBβ1 overexpression moderately increased 12-O-tetradecanoylphorbol-13-acetate- and interleukin-1-inducible NF-κB DNA binding. 12-O-Tetradecanoylphorbol-13-acetate- and interleukin-1-induced transactivation by NF-κB, however, was lower in IκBβ1-overexpressing cells. Mutants of IκBβ1 lacking the C-terminal casein kinase II phosphorylation sites, which form a stable complex with DNA bound NF-κB without inhibiting its transactivation in other cell types, repressed the transactivation by NF-κB in MDA-MB-231 cells. Consistent with the results of transient transfections, the expression of urokinase plasminogen activator, an NF-κB target gene, was reduced in IκBβ1-overexpressing cells. These results suggest that depending on the cell type, IκBβ1 represses the expression of NF-κB-regulated genes by inhibiting either DNA binding or transactivation function of NF-κB.

Regulation of gene expression by the NF-B 1 /Rel family of transcription factors is controlled mainly by the inhibitory IB proteins which include IB␣, IB␤, IB␥, and IB⑀ (1)(2)(3)(4). The active complex of NF-B is composed of homodimers and heterodimers of p50, RelA, RelB, and c-Rel. These complexes are sequestered in the cytoplasm by IBs. Extracellular signalinduced phosphorylation and subsequent degradation of IBs is essential for nuclear translocation of NF-B. NF-B binds to the recognition elements in the promoter region of target genes and activates transcription. Transactivation by NF-B involves interaction with transcriptional coactivators such as CBP/p300 and SRC-1 and general transcription factors including TBP (5)(6)(7)(8). In addition, NF-B interacts with transcription factors such as activator protein 1 (AP-1) and serum response factors and synergistically activates transcription (9,10).
IB proteins determine the duration of transactivation by NF-B. IB␣ is involved in transient activation of NF-B because it is degraded rapidly upon stimulation and is resynthesized by activated NF-B (1)(2)(3)(4). Most of the inducers of NF-B cause degradation of IB␣. The newly synthesized IB␣ sequesters NF-B in the cytoplasm and terminates the signal. IB␤ is involved in persistent activation of NF-B (11). Degradation of IB␤ is much more delayed than IB␣ and resynthesis is independent of NF-B (11). Degradation of IB␤ is dependent on the cell type and the extracellular signals that induce NF-B. For example, whereas interleukin 1 (IL-1) induces degradation of IB␤ in most cell types tested, tumor necrosis factor (TNF-␣) induces degradation of IB␤ in E29.1 T cell hybridomas but not in Jurkat cells (11,12). Together, but not separately, TNF-␣ and interferon ␥ induce degradation of IB␤ in PC12 cells (13). The ability of IB␤ to repress NF-B DNA binding is determined by the basal phosphorylation of the C-terminal PEST domain (14). The newly synthesized hypophosphorylated IB␤ functions as a chaperone for NF-B by protecting it from IB␣ (15). The IB␤⅐NF-B complex enters the nucleus, binds to DNA, and activates transcription (16,17). The phosphorylation of the PEST domain by casein kinase II or the association of the IB␤⅐NF-B complex with high mobility group I (HMG I) proteins on selected promoters converts IB␤ to a repressor of NF-B DNA binding activity (14,17).
Two isoforms of IB␤, ␤1 and ␤2, have been described recently (18). These isoforms differ in their C-terminal amino acids. IB␤1 is functionally similar to IB␣ and efficiently represses the activity of p50:RelA heterodimers. It is found in both the nucleus and cytoplasm and is degraded by various stimuli. In contrast, IB␤2 is a cytoplasmic protein that is refractory to signal-induced degradation. Although both isoforms are indistinguishable in their binding preferences to cellular NF-B/Rel homo-and heterodimers, IB␤2 is more effective in inhibiting RelA homodimers than p50:RelA and p50:c-Rel heterodimers (18). This property of IB␤2 is similar to that of IB⑀ which preferentially inhibits the activity of RelA homodimers (19).
There is increasing evidence for the role of NF-B in cancer progression and resistance to chemotherapy (20,21). Constitutive activation of NF-B has been observed in Hodgkin's lymphoma, melanomas, juvenile myelomonocytic leukemia, cuta-neous T cell lymphoma, and human T cell leukemia virus type I Tax protein-mediated cellular transformation (22)(23)(24)(25)(26). We and others (27,28) have shown that NF-B is constitutively active in a subset of breast cancers. In this study, we have investigated the role of IB␣ and IB␤ in constitutive activation of NF-B in breast cancer cells. Constitutive DNA binding correlated with reduced levels of either IB␣ or IB␤. Both isoforms of IB␤ are present in breast cancer cell lines and primary breast cancers. Overexpressed IB␣ but not IB␤1 repressed constitutive DNA binding of NF-B in MDA-MB-231 cells. IB␤1, in contrast, repressed transactivation by NF-B. Unlike in the mouse embryonal carcinoma cell line P19 and the human embryonic kidney cell line 293, mutants of IB␤1 that cannot be phosphorylated by casein kinase II efficiently repressed transactivation by NF-B in MDA-MB-231 cells (17,29). Taken together, these results reveal a novel cell typespecific inhibitory regulation of transactivation function but not DNA binding of NF-B by IB␤1.

EXPERIMENTAL PROCEDURES
Cell Culture-All breast cancer cell lines were purchased from ATCC, and their growth conditions have been described previously (30). 10 Ϫ9 M insulin was added to MDA-MB-231 and its clones during long term culturing.
Recombinant Plasmids-NF-B/CAT reporter gene, human IB␣, and mouse IB␤ expression vectors have been described previously (27). The AP-1/CAT reporter was a generous gift of E. O'Neill (Merck) and contains three copies of an AP-1 site from the human metallothionein gene (31). Human IB␤ and its mutants were generous gifts from D. Ballard (Vanderbilt University, Nashville, TN) and were described previously (29). mIB␤1⌬C, which lacks 50 amino acids at the Cterminal, was generated by cloning a NotI-HindIII fragment of mIB␤1 to the pcDNA3 expression vector (Invitrogen). Retrovirus constructs containing hIB␣ and mIB␤1 coding sequences were prepared by cloning cDNAs into the respective EcoRI and NotI sites of the modified retrovirus vector LxSN (32).
Retrovirus Preparation and Transduction-LxSN vectors were amphotrophically packaged into the AM12 packaging cell line (33), and the cell-free supernatant was used to transduce MDA-MB-231 cells as described previously (34). Vector containing clones were isolated by growing cells in G418 (1 mg/ml).
Electrophoretic Mobility Shift Assay (EMSA)-Whole cell extracts from cells were prepared and subjected to EMSA as described previously (27). Densitometric analysis was performed to classify constitutive DNA binding activity.
Western Blotting-Cell extracts were prepared in radioimmunoprecipitation assay buffer (35) and subjected to Western blotting as recommended by the antibody manufacturers (Santa Cruz Biotechnology and Sigma). For quantitation, autoradiograms were scanned using a densitometer.
DNA Immunoprecipitation-Nuclear extracts were prepared as described previously (36). Nuclear extracts supplemented with 1% Nonidet P-40 were precleared in insoluble protein A solution (Sigma) and subjected to EMSA reaction. Antibodies (0.2 g) were added, and incubation was carried out for 1 h on ice. Protein A-agarose (20-l bead volume) was added, and the reaction mixture was incubated for 1 h at 4°C with gentle rocking. Protein A-agarose was collected by centrifugation, washed five times in 1ϫ EMSA buffer (10 mM Tris-HCl, 0.5 mM EDTA, 100 mM KCl, 10% glycerol, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 0.2% Nonidet P-40), and subjected to SDS-polyacrylamide gel electrophoresis. The amount of radioactive DNA immunoprecipitated with the antibody complex was visualized by autoradiography and quantitated by densitometric scanning.
Transient Transfections and CAT Assays-Transient transfections and the CAT assay have been described previously (27). Statistical analysis was performed using Statview software ( Concepts, Berkeley, CA). Data were evaluated by one-factor analysis of variance and Fisher's protected least significant difference as a post hoc test.
Serum and Epidermal Growth Factor (EGF) Treatment and Northern Blotting-For serum stimulation, cells were maintained in MEM without serum for 48 h. MEM with 10% fetal calf serum was added, and the cells were harvested after 4 h. For EGF treatment, cells were maintained in MEM (without fetal calf serum) for 48 h. EGF (30 ng/ml) was added, and cells were harvested after 4 h. RNA was prepared by the guanidinium isothiocyanate/cesium chloride method and subjected to Northern blotting as described previously (30).

Differential Expression of IBs in Breast
Cancer Cells-Constitutive NF-B DNA binding activity in several breast cancer cell lines was compared with the expression level of IB␣ and IB␤ ( Fig. 1 and Table I) Table I). Among these cell lines, MCF-7, T47-D, and ZR-75-1 express estrogen receptor ␣ (ER␣) and are more differentiated than other cell lines (37). Most of the cell lines that displayed elevated NF-B DNA binding activity are ER␣-negative, consistent with our previous report (27). In contrast to NF-B, DNA binding of SP1 and CTF/NF-1 family transcription factors did not correlate with ER␣ status (Fig. 1A). The NF-B complex is a heterodimer of RelA and p50 subunits as determined by antibody supershift assays (data not shown). In general, NF-B DNA binding activity correlated inversely with the levels of IB␣, IB␤1, and IB␤2 (Fig. 1B, Table I IB␤2 is the major IB␤ in breast cancer cell lines with the exception of HBL-100 and MDA-MB-436 cells which contained similar levels of IB␤1 and IB␤2. Reverse transcription-polymerase chain reaction was used to confirm the presence of transcripts corresponding to IB␤1 and IB␤2 in breast cancer cells (data not shown). Proteins corresponding to IB␤1 and IB␤2 were also observed in primary breast cancers (data not shown). There was no correlation between IB expression and NF-B DNA binding activity in Hs578T and HBL-100 cells. Despite reduced levels of both IB␣ and IB␤2, NF-B DNA binding was intermediate in Hs 578T. It is possible that this cell line overexpresses other IB proteins. HBL-100 cells displayed elevated NF-B DNA binding activity despite the presence of IB␣ and IB␤2. Whether elevated NF-B DNA binding in this cell line is due to 1:1 ratio of IB␤1 and IB␤2 or due to reduced levels of other IB proteins is not known. Constitutive NF-B DNA binding activity did not correlate with elevated levels of RelA and p50 proteins (Fig. 1B).
The Effect of IB␣ and IB␤1 Overexpression on Constitutive NF-B DNA Activity in MDA-MB-231 Cells-To determine the role of IB␣ and IB␤ in constitutive NF-B activation, we generated MDA-MB-231 cells overexpressing IB␣ and IB␤1. The levels of IB␣ and IB␤1 in cells transduced with empty retrovirus (LxSN3), IB␣ coding retrovirus (IB␣3-7), and IB␤1 containing retrovirus (IB␤21-25) were determined by Western blotting (Fig. 2A). Two clones containing vector alone (called LxSN3 and LxSN5 cells hereafter), two clones with IB␣ (IB␣5 and IB␣7 cells), and three clones with IB␤1 (IB␤21, IB␤22, and IB␤23) were chosen arbitrarily for this study. NF-B DNA binding activity in these clones is shown in Fig. 2B. IB␣ overexpression resulted in reduced constitutive DNA binding (compare lane 1 and 2 with lanes 3 and 4). Unexpectedly, constitutive NF-B DNA binding was the same in IB␤ cells and LxSN cells (compare lanes 1 and 2 with lanes 5-7). Similar results were obtained when nuclear extracts instead of whole cell extracts were used (Fig. 2B, lanes 8 -14). Also, a similar binding pattern was obtained with PRDII element (NF-B-binding site in the interferon-␤ gene promoter which is flanked by the HMG I protein-binding site) and the NF-B-binding site of urokinase plasminogen activator (uPA) promoter (data not shown). The DNA⅐NF-B complex is a heterodimer of p50 and RelA as antibodies against both proteins supershifted the complex (data not shown). IB␣ or IB␤ overexpression did not affect DNA binding of AP-1 (Fig. 2B, lower  panel).
As stated earlier, the hypophosphorylated IB␤1 enters the nucleus and forms a stable complex with DNA-bound NF-B. Inability of IB␤1 to reduce NF-B DNA binding activity could be due to unimpaired interaction of the IB␤1⅐NF-B complex with DNA. To address this possibility, we performed DNA immunoprecipitation assays with nuclear extracts from LxSN3, IB␣5, and IB␤21 cells. Radiolabeled DNA⅐NF-B complexes were immunoprecipitated with either RelA antibody, IB␣ antibody, or IB␤ antibody. Antibody against neu oncogene was used as a negative control. DNA⅐NF-B complex was efficiently immunoprecipitated by RelA antibody (Fig. 2C,  lanes 1, 5, and 9). The amount of DNA⅐NF-B complex immunoprecipitated by RelA antibody was similar to the amount of NF-B⅐DNA complex identified in EMSA. Neither IB␣ antibody nor neu antibody immunoprecipitated DNA⅐NF-B complex from nuclear extracts (lanes 2, 4, 6, 8, 10, and 12). IB␤ antibody immunoprecipitated radiolabeled probe from only IB␤21 cells (compare lanes 3, 7, and 11). These results indicate that IB␤1 either binds directly to the NF-B-binding site or forms a stable complex with DNA-bound NF-B. Whatever the mechanism involved, DNA binding of IB␤1 did not interfere in DNA-NF-B interaction. IB␤1-DNA interaction was TABLE I Summary of NF-B DNA binding activity and IB expression in breast cancer cell lines ER␣ status in these cell lines has been described previously (30,37). A value of 4ϩ is assigned to the most intense signal among various cell lines.
Ϫ ϩϩ ϩ ϩϩ ϩ HBL100 Ϫ ϩϩϩϩ ϩϩϩϩ ϩϩϩ ϩϩϩ not observed in antibody supershift assays possibly due to the unstable nature of the complex under electrophoresis conditions (data not shown).
Inducible DNA Binding of NF-B in Cells Overexpressing IB␣ and IB␤-To investigate the effect of IB␣ and IB␤ on inducible DNA binding of NF-B, we performed EMSA with extracts from cells treated with either TPA or IL-1␤ for 1 h. Inducible NF-B DNA binding activity was lower in IB␣ cells compared with other cells (Fig. 3A). As with constitutive DNA binding, IB␤ failed to inhibit inducible NF-B DNA binding. In fact, we consistently observed higher DNA binding in TPAtreated IB␤ cells compared with other cells (compare lanes 2, 5, 14, 17, and 20). IB␣ and IB␤ overexpression did not affect DNA binding of AP-1 (Fig. 3A, lower panel).
To determine whether IB␣ and IB␤ overexpression alters the time course of NF-B activation, EMSA was performed with the extracts from cells treated with TPA and IL-1 for specific intervals. TPA-inducible DNA binding activity was observed after 60 min of stimulation in all three cell types (Fig.  3B, compare lanes 4, 13, and 22) and started to decline after 3 h of stimulation. We consistently observed increased NF-B DNA binding activity in TPA-treated IB␤ cells compared with other cells. There was no cell type-specific variation in the rate of IB␣ degradation. Approximately 50% of IB␣ was present at 60 min and 2 h after TPA addition, and its level increased at 3 h (Fig. 3B). NF-B DNA binding activity at 6, 9, and 24 h after TPA addition was higher than in untreated cells which correlated with reduced levels of IB␣. Degradation of IB␤1 was minimal (Ͻ25%) which was evident at 2 and 3 h post-stimulation. We next examined the time course of IL-1-induced NF-B DNA binding activity (Fig. 3C). NF-B DNA binding increased within 30 min of IL-1 addition in all three cell types. There was a difference in the termination of inducible DNA binding. In LxSN5 and IB␣5 cells, maximum DNA binding was observed after 1 h of IL-1 addition and returned almost to basal levels by 2 h. In contrast, in IB␤21 cells, a substantial level of inducible DNA binding persisted after 2 h of IL-1 addition. Similar results were obtained with three other IB␤-overexpressing clones (data not shown). As with TPA-treated cells, IB␣ degradation occurred at a similar rate in all three cell types, and partial degradation of IB␤ (Ͻ30%) coincided with induction of NF-B DNA binding in IB␤21 cells (data not shown). Taken together, our results demonstrate that although IB␣ did inhibit constitutive DNA binding, neither IB␣ nor IB␤1 had much effect on inducible DNA binding of NF-B in MDA-MB-231 cells.
Activity of NF-B-and AP-1-dependent Promoters in IB␣and IB␤1-overexpressing Cells-The inability of IB␤1 to repress constitutive DNA binding of NF-B was unexpected. However, it is possible that IB␤1 may convert the NF-B⅐DNA complex to a transcriptionally inactive complex similar to retinoblastoma protein-induced inactivation of the NF-B⅐DNA complex (38). To address this possibility, we performed transient transfection experiments using an NF-B-dependent CAT reporter gene (NF-B/CAT, see Ref. 27). Expectedly, constitutive and inducible activities of the NF-B/CAT were lower in IB␣ cells compared with LxSN cells (Fig. 4A). Interestingly, inducible activities of NF-B/CAT, particularly with IL-1, were lower in IB␤ cells. The NF-B/CAT reporter gene expression requires the binding site for NF-B because a similar reporter in which the NF-B-binding site has been mutated is inactive in MDA-MB-231 cells (27).
As stated in the Introduction, NF-B interacts with AP-1, and this interaction results in synergistic activation of transcription. Because IB␤1 inhibited NF-B-dependent gene expression despite efficient DNA binding, it is possible that IB␤1 in the NF-B⅐DNA complex prevents synergistic interaction with AP-1 either due to a conformational change or steric hindrance. We measured constitutive and TPA-inducible activity of AP-1/CAT reporter gene in all cell lines. Similar levels of constitutive and TPA-inducible activity were observed in LxSN5, IB␣5, and IB␣7 cells (Fig. 4B). Due to differences in cell passage number, we consistently observed lower constitutive and TPA-inducible activities in LxSN3 cells. These cells were passed ϳ15 more times than other clones. For unknown reasons, decreasing levels of AP-1 transcriptional activation with increasing cell passage were consistently observed with this cell line. The constitutive and TPA-inducible activities of  (bottom panel). C, IB␤1 is associated with NF-B-binding site containing probe. An EMSA reaction with radiolabeled NF-B probe was performed with 15 g of nuclear extracts from indicated clones. EMSA reactions were subjected to immunoprecipitation using indicated antibodies. Antibodies against RelA and neu served as positive and negative controls, respectively. The amount of radioactive probe precipitated by RelA antibody from LxSN3 cells was set as 100%, and the relative difference is indicated.
AP-1 were reproducibly lower (n ϭ 6) in IB␤ cells compared with other clones. Similar results were obtained with interleukin-11 (IL-11) promoter/CAT reporter with an AP-1-binding site (data not shown). The activity of the reporter gene is AP-1-dependent since an IL-11/CAT reporter in which the AP-1-binding site is mutated was inactive (data not shown). To confirm these results further, we performed transient transfection assays with AP-1/CAT reporter, RelA, and IB expression vectors in MDA-MB-231 cells. Because of a higher level of basal AP-1 activity, only a modest increase in AP-1/CAT activity in RelA-transfected cells was observed (Fig. 4C) (17,29). Alternatively, promoter context-dependent interaction of HMG I with the IB␤⅐NF-B⅐DNA complex results in the release of NF-B and IB␤ from DNA (17). Neither of the above two mechanisms appears to be responsible for reduced NF-B/CAT activity in IB␤1-overexpressing MDA-MB-231 cells because transactivation function but not the DNA binding of NF-B was inhibited by IB␤1 in these cells. To verify this further, we compared the ability of wild type and mutants of IB␤ (in which casein kinase II phosphorylation sites have been mutated to alanine or aspartic acid) to repress NF-B DNA binding activity in MDA-MB-231 and 293 cells. The effect of IB␤1 expression vector lacking 50 amino acids at the C terminus (mIB␤1⌬C which lacks casein kinase II phosphorylation sites) on NF-B DNA . IB␣ in all cell types and IB␤1 in IB␤21 cells were analyzed by Western blotting. Western blot with LxSN5 cell extracts was exposed three times longer than other blots, and it was performed with double the amount of proteins. As a loading control, the same blots were reprobed with an antibody against ␣-tubulin. C, time course of IL-1-inducible NF-B DNA binding activity. The experiments were performed as in B.
binding activity was also tested in parallel. Wild type but not the mutants of IB␤1 repressed NF-B DNA binding activity in 293 cells, which is consistent with previously published data (Fig. 5A). In contrast, both wild type and mutant IB␤1 failed to repress NF-B DNA binding in MDA-MB-231 cells. The C-terminal serines of IB␤1 appear to be phosphorylated in both cell types as the mutant proteins with either serine to alanine or serine to aspartic acid substitution displayed faster mobility than wild type IB␤1 on an SDS-polyacrylamide gel electrophoresis (Fig. 5A). We next compared the ability of wild type and mutant IB␤1 to repress transactivation function of NF-B in transient transfection assays. Mutants were as efficient as wild type IB␤1 in repressing NF-B activity (Fig. 5B). MDA-MB-231 cells stably expressing mIB␤1⌬C were generated to confirm further these results (Fig. 5C). A new clone containing vector alone was also prepared so that all clones were of the same passage number. mIB␤1⌬C overexpression did not affect constitutive NF-B DNA binding activity (Fig.  5C). As with the transient transfection assays, however, NF-B/CAT activity was reduced in clones overexpressing IB␤1⌬C (Fig. 5D). Taken together, these results suggest that phosphorylation of IB␤1 by casein kinase II is not essential for repression of NF-B activity in MDA-MB-231 cells.
Serum and EGF-inducible Expression of uPA in IB␣ and IB␤-overexpressing Cells-Multiple transcription factors including AP-1, NF-B, and the Ets are involved in constitutive and inducible expression of the uPA gene (39,40). If the effect of IB␤ on NF-B-and AP-1-dependent gene expression observed in transient assays is relevant to endogenous gene expression, IB␤ overexpression should inhibit uPA gene expression. Serum and EGF were used as inducible agents as both can induce the activity of AP-1 and NF-B (41,42). The basal expression level of uPA was consistently lower in LxSN3 compared with LxSN5, possibly due to the difference in passage number (Fig. 6). Both constitutive and inducible expression of uPA was reproducibly lower in IB␣ and IB␤ cells (Fig. 6, A  and B). In general, the inducible expression was lower in IB␤ cells compared with IB␣ cells. Similar results were also obtained with cells overexpressing IB␤1⌬C (Fig. 6C). The effect of EGF and serum on AP-1 and NF-B is mostly at the level of transactivation as DNA binding of both transcription factors was not significantly affected upon EGF and serum treatment (data not shown). These results further emphasize that IB␤1 can repress transcription without interfering with the DNA binding of NF-B in MDA-MB-231 cells.
uPA Expression in Breast Cancer Cell Lines-The above results indicate that NF-B plays an important role in uPA expression in breast cancer cells. To extend this observation further, we compared uPA expression with constitutive NF-B DNA binding activity in various breast cancer cell lines. uPA expression was observed only in cell lines that contained either intermediate or higher levels of constitutive NF-B activity (Fig. 7). Interestingly, uPA expression in three out of five cell lines correlated with reduced levels of IB␤ (MDA-MB-231, MDA-MB-436, and Hs578T). These results suggest that for optimum activation of NF-B-regulated genes, constitutive NF-B DNA binding should be accompanied by reduced levels of IB␤1.

DISCUSSION
Members of the Rel family of transcription factors are usually sequestered in the cytoplasm by IB proteins. Several recent studies have indicated that IB-mediated cytoplasmic retention of NF-B is altered during cancer progression. Although reduced levels of IB proteins may be responsible for nuclear NF-B in some cells, constitutive NF-B DNA binding has also been observed under conditions where IB␣ is abun- dant (24,43). Tyrosine phosphorylation-mediated inactivation without degradation of IB␣ or accumulation of hypophosphorylated IB␤ which transports NF-B to the nucleus may account for constitutive NF-B DNA binding in cells with IB␣ (15,43). This study was initiated to understand the mechanisms of constitutive NF-B activation in breast cancer cells. have shown that the strength by which different IBs inhibit NF-B activity correlates with their ability to sequester NF-B in the cytoplasm rather than their ability to inhibit DNA binding of NF-B. They have also shown that the first three of six ankarin repeats determine the cytoplasmic retention potential of IBs, whereas the last three ankarin repeats and the Cterminal PEST sequences determine the ability of IBs to inhibit DNA binding. IB␤1 could neither prevent nuclear translocation nor prevent DNA binding of NF-B in MDA-MB-231 cells ( Fig. 2 and Fig. 5). IB␤1 was, however, able to inhibit transactivation by NF-B in these cells. Phosphorylation of PEST domain of IB␤1 by casein kinase II (which is required for inhibition of DNA binding (16,17)) is not necessary for repression of transactivation function of NF-B by IB␤1 in MDA-MB-231 cells. We observed similar degrees of repression of NF-B activity by wild type and mutant IB␤1 lacking casein kinase II phosphorylation sites (Fig. 5). These results suggest that IB␤1 contains additional functional domains which only inhibit the transactivation function of NF-B. Detailed mutational analysis and domain swapping with other IBs is essential for characterization of this domain. We have not directly tested whether this function of IB␤1 is promoter context-dependent. In this regard, note that IB␤1 reduced the expression of uPA and the transiently transfected reporter gene but failed to prevent the resynthesis of IB␣ (which is dependent on NF-B) in cells treated with TPA ( Fig. 3 and Fig.  6). Because the NF-B-binding site in the reporter gene and uPA gene lacks the binding sites for HMG I, the repression of NF-B activity is not likely due to the interaction of HMG I with NF-B-IB␤-DNA. IB␤ may inhibit transactivation by NF-B independent of its interaction with NF-B or through its association with DNA-bound NF-B. IB␤1 may sequester a coactivator(s) which is essential for transactivation by NF-B. In this regard, proteins distinct from the IB/NF-B family have been shown to interact with IB␤1 (45,46). IB␤1 in the NF-B⅐DNA complex may inhibit transactivation in the following ways. (i) IB␤1 may mask the region of RelA which undergoes a modification such as phosphorylation after DNA binding. Several recent studies have suggested that RelA phosphorylation is required for efficient transactivation. For example, Bergmann et al. (47) have demonstrated that inhibitors of phosphatidylcholine-specific phospholipase C and protein kinase C blocked transactivation by NF-B after TNF and IL-1␤ treatment even though these inhibitors did not block nuclear translocation. Similarly, Yoza et al. (48) have demonstrated that activation of the protein-tyrosine kinase pathway is required for lipopolysaccharide-mediated NF-B activation but not nuclear translocation. Phosphorylation of RelA by kinases activated by TNF-␣ and ras oncogene has also been reported (49,50). RelA appears to be the target of protein kinase A or casein kinase II following degradation of IBs (51,52). Protein kinase A-mediated phosphorylation by RelA is required for unmasking of intramolecular interaction sites and subsequent interaction of RelA with FIG. 6. Effect of IB␣ and IB␤ overexpression on uPA expression. A, serum-inducible expression of uPA. Total RNA (20 g) from indicated clones maintained in serum-free medium for 48 h (Ϫ) or treated for 4 h with 10% serum-containing media after serum starvation (ϩ) was subjected to Northern blotting using a uPA probe. Integrity of the RNA was examined by reprobing the blot with ribosomal protein gene 36B4 (30). The ratio between uPA and 36B4 was determined by densitometric scanning of autoradiograms. Results from a representative experiment are shown. B, EGF-inducible expression of uPA. Total RNA from untreated (Ϫ) or EGF-treated (30 ng/ml for 4 h) was subjected to Northern blotting as above. C, serum and EGF-inducible expression of uPA in cells overexpressing mIB␤1⌬C. coactivator p300/CBP (53). It is possible that phosphorylation by protein kinase A and other kinases and/or coactivator-NF-B interaction is inefficient when IB␤1 is present in the NF-B⅐DNA complex. (ii) IB␤ has been shown to interact with nuclear receptors including thyroid hormone receptors and retinoid-X receptors (45,46). Because nuclear receptors such as estrogen receptor, progesterone receptors, and peroxisome proliferator receptors have been shown to inhibit NF-B activity, IB␤1 in the NF-B⅐DNA complex may recruit retinoid-X receptors:peroxisome proliferator receptor heterodimers to the NF-B⅐DNA complex and inhibit transactivation by NF-B (27, 54 -56). (iii) IB␤1 may inhibit synergistic interaction of NF-B with other transcription factors including AP-1. We have provided some evidence for the last possibility. Synergistic crosstalk between NF-B and AP-1 is essential for anchorage-independent growth of immortalized keratinocytes (57). Inhibitors of both NF-B and AP-1 block tumor promoter-induced transformation (58), suggesting the importance of cross-talk in cancer progression. We suspect that such cross-talk is more pronounced in cancer cells lacking IB␤1. These cells will be more responsive to growth factors and serum. Consistent with this possibility, serum and EGF-inducible expression of uPA was more pronounced in cells lacking IB␤1. We have observed increased uPA expression in most of the breast cancer cell lines that contain constitutively active NF-B. uPA is absolutely required for intravasation, the most important step in the multistep process of tumor metastasis (59), and its overexpression in breast cancer cells is associated with poor prognosis (60). In summation, our results suggest that IB␤ regulates NF-B activity in a cell type-specific manner which determines whether constitutive DNA binding of NF-B leads to overexpression of NF-B-regulated genes such as uPA.