Differential IκB Kinase Activation and IκBα Degradation by Interleukin-1β and Tumor Necrosis Factor-α in Human U937 Monocytic Cells

The IκB kinases (IKKs) lie downstream of the NF-κB-inducing kinase (NIK) and activate NF-κB by phosphorylation of IκBα. This leads to IκBα degradation and release of NF-κB. In U937 monocytic cells, interleukin (IL)-1β (1 ng/ml) and tumor necrosis factor (TNF)-α; 10 ng/ml) induced κB-dependent transcription equally. However, IKK activity was strongly induced by TNF-α but not by IL-1β. This was consistent with IκBα phosphorylation and degradation, yet TNF-α-induced NF-κB DNA binding was only 30–40% greater than for IL-1β. This was not explained by degradation of IκBβ, IκBε, or p105 nor nuclear translocation of NF-κB·IκBα complexes or degradation-independent release of NF-κB. Dominant negative (NIK) repressed TNF-α and IL-1β-induced κB-dependent transcription by ∼60% and ∼35%, respectively. These data reveal an imprecise relationship between IKK activation, IκBα degradation, and NF-κB DNA binding, suggesting the existence of additional mechanisms that regulate NF-κB activation. Finally, the lack of correlation between DNA binding and transcriptional activation plus the fact that PP1 and genistein both inhibited κB-dependent transcription without affecting DNA binding activity demonstrate the existence of regulatory steps downstream of NF-κB DNA binding. Therapeutically these data are important as inhibition of the NIK-IKK-IκBα cascade may not produce equivalent reductions in NF-κB-dependent gene expression.

The acute phase transcription factor nuclear factor B (NF-B) 1 is an inducible enhancer of many inflammatory genes including cytokines, chemokines, and adhesion molecules as well as enzymes such as inducible nitric-oxide synthase and cyclooxygenase-2 (reviewed in Refs. 1 and 2). Proinflammatory cytokines such as interleukin (IL)-1␤ and tumor necrosis factor (TNF)-␣ rapidly induce NF-B DNA binding and B-dependent transcription in most cell types (1,2). NF-B DNA binding activity comprises homo-and, more usually, heterodimers of Rel proteins such as RelA (p65), RelB, c-Rel, NF-B1 (p105/ p50), and NF-B2 (p100/p52) (2). NF-B is held inactive in the cytoplasm by inhibitory IB proteins including IB␣, IB␤, IB␥, and IB⑀. Agonists such as TNF-␣ and IL-1␤ result in phosphorylation of IB␣ at serines 32 and 36 (3,4). These phosphorylation events lead to ubiquitination of IB␣ followed by its rapid degradation by the 26 S proteasome (5). This releases active NF-B, typically p50/p65 heterodimers, which translocates to the nucleus and activates transcription via B enhancer elements. Recently a high molecular mass, Ϸ700 kDa, complex has been described that contains kinase activities specific for serines 32 and 36 of IB␣ (6 -8). The two main IB␣ kinase (IKK) activities in this complex, termed the IKK signalsome, have been cloned and are called IKK␣ and IKK␤. In addition, the upstream kinase, where the IL-1␤ and TNF-␣ signaling pathways converge before IKK activation, has been identified as a mitogen-activated protein kinase kinase kinase and named NF-B-inducing kinase (NIK) (9). As NIK interacts with and stimulates IKK activity and can phosphorylate IKK␣ on serine 176, it is likely that NIK lies immediately upstream of the IKKs in the NF-B activation cascade (10 -12).
However, the degree to which these signaling events apply to different cell types and different inducing agents is presently unclear. In asthmatic individuals, elevated levels of activated NF-B are found in sputum macrophage, suggesting that cells of the monocytic lineage may play a significant role in asthmatic inflammation (13). Human monocytic U937 cells were therefore used to investigate the NIK-IKK-IB signal transduction pathway in the activation of NF-B and B-dependent transcription by IL-1␤ and TNF-␣.

EXPERIMENTAL PROCEDURES
Cell Culture-U937 cells (ECACC code 85011440) were cultured at 37°C in a humidified atmosphere with 5% CO 2 in RPMI 1640 medium (Sigma) supplemented with 10% (v/v) fetal calf serum (Sigma), 2 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 2.8 g/ml amphotericin B (complete medium) and were maintained between 2-9 ϫ 10 5 cells/ml. Cells were stimulated in RPMI medium as above but supplemented with 2% fetal calf serum at 5 ϫ 10 5 cells/ml. IL-1␤ and TNF-␣ (R&D Systems, Abingdon, Oxon) were used at 1 ng/ml and 10 ng/ml, respectively. Where used, cycloheximide (10 g/ml; Sigma) was added 5 min before stimulation, N-carbobenzyloxy-Leu-Leu-leucinal (MG-132) (10 M; Sigma) was added 60 min before stimulation, and PP1 and genistein (Calbiochem) were added 30 min before stimulation at concentrations of 10 M and 100 M, respectively. Drugs were dissolved in dimethyl sulfoxide (Me 2 SO) and were diluted to final concentrations of less than 0.1% (v/v). At this level Me 2 SO had no effect on activation of NF-B or B-dependent transcription (data not shown).
Plasmids-The NF-B-dependent reporter, pGL3.6B.BG.luc, contains two tandem repeats of the sequence 5Ј-GGG GAC TTT C CC TGG GGA CTT TCC CTG GGG ACT TTC CC-3Ј, which contains three copies of the decameric NF-B binding site (underlined) upstream of a minimal ␤-globin promoter driving a luciferase gene as described previously (14). The reporter, pGL3.6Bmut.luc, is as above except that the core NF-B binding site is mutated to 5Ј-GCC ACT TTC C-3Ј (mutated bases underlined). The NIK and dominant negative NIK expression vectors were gifts from David Wallach (9).
Transient Transfections-Aliquots of 10 ϫ 10 6 cells in 250 l of Hanks' balanced salt solution were electroporated with 10 g of either pGL3.6B.BG.luc or pGL3.6B.BG.luc using a Gene Pulser II (Bio-Rad) set for 200 V, 950 microfarads. For overexpression studies, electroporation was performed using either 5 g of pGL3.6B.BG.luc plus 5 g of pcDNA3 or NIK expression vector or using 2.5 g of pGL3.6B.BG.luc plus 10 g of pcDNA3 or dominant negative NIK expression vector. Transfected cells were incubated in 10 ml of complete medium for 12 h before plating onto 2ϫ 6-well plates in RPMI medium containing 2% fetal calf serum. Cells were harvested 6 h after stimulation and assayed for luciferase activity using a commercially available luciferase reporter gene assay (Promega, Southampton, UK). After normalization to protein concentration, data were expressed as fold activation.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear proteins, 5 g, were used in binding reactions as described previously (16). Consensus NF-B probe (Promega) containing the decameric NF-B site (underlined) was 5Ј-AGT TGA GGG GAC TTT CCC AGG-3Ј (sense stand). Specificity was determined by the prior addition of a 100-fold excess of unlabeled competitor consensus oligonucleotide. For supershift analysis, nuclear extracts were incubated on ice for 90 min with antisera raised to various Rel proteins (Santa Cruz Biotechnology, Santa Cruz, CA) at 0.4 g/ml before the addition of radiolabeled oligonucleotide. Reactions were separated on 6% nondenaturing acrylamide gels in 0.25 ϫ Tris-buffered EDTA. Gels were dried, and protein-DNA complexes were visualized by autoradiography.
Immunoprecipitation-Cytoplasmic extracts, 100 g, prepared in Buffer A supplemented with 25 g/ml aprotinin, 10 g/ml leupeptin, 0.5 mM Na 3 VO 4 , 1 mM sodium pyrophosphate, 50 mM NaF, and 5 g/ml N-tosyl-L-phenylalanine chloromethyl ketone (TPCK) were precleared by the addition of 1.0 g of normal rabbit IgG (Santa Cruz) together with 20 l of protein A-Agarose (Santa Cruz) for 1 h at 4°C. After centrifugation, supernatants were incubated with 5 l of p65 antibodyagarose conjugate (SC-109AC) (Santa Cruz) for 2 h at 4°C, and immunoprecipitates were collected by centrifugation. After washing 4 times with buffer A supplemented with inhibitors and 0.05% Tween 20, immunocomplexes were resuspended in 40 l of 1ϫ SDS loading buffer and boiled for 5 min before Western blot analysis.
Metabolic Labeling of p65-Cells were incubated in phosphate-free RPMI media for 2 h before the addition of [ 32 P]orthophosphate (Amersham Pharmacia Biotech) at 0.2 mCi/ml and incubated for a further 4 h. Cells were then treated as indicated before harvesting in 200 l of 1ϫ phosphate-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS (radioimmune precipitation buffer) supplemented with 0.5 mM PMSF, 2 mM Na 3 VO 4 , 10 g/ml leupeptin, 25 g/ml aprotinin, and 50 mM NaF on ice for 15 min. Immunoprecipitation of p65 was performed as above except that radioimmune precipitation buffer was used in place of Buffer A. Samples were run on 10% SDS-PAGE before autoradiography or Western blotting.
EMSA revealed that both IL-1␤ and TNF-␣ strongly induced NF-B DNA binding (Fig. 1B). However, the TNF-␣ response occurred with faster kinetics than the IL-1␤ response and generally showed the highest overall activation (Fig. 1B). Thus TNF-␣-induced NF-B was near maximal by 15 min, whereas IL-1␤-induced NF-B was not maximal until 30 min post-stimulation. Supershift analysis was performed to identify the Rel proteins involved in these responses (Fig. 1C). In IL-1␤ and TNF-␣ treated extracts, anti-p50 antisera and anti-p65 antisera resulted in reduced mobility of DNA binding complexes. The NF-B complex was unaffected by anti-c-Rel, RelB, and p52 antisera. These data indicate that in U937 cells, IL-1␤and TNF-␣-inducible NF-B DNA binding complexes are made up of both p50 and p65 proteins and not c-Rel, RelB, or p52. The fact that heterogeneous bands were observed suggests that both homo-and heterodimers of p50 and p65 may be present.
IB Degradation by IL-1␤ and TNF-␣-Western blot analysis was performed on cytoplasmic extracts to examine the effect of IL-1␤ and TNF-␣ on IB␣ degradation. Cells treated with vehicle showed no change in IB␣ protein for the duration of the experiment ( Fig. 2A, left panel). After IL-1␤ treatment, little or no change in IB␣ levels was observed during the first 15 min post-stimulation, and by 30 min, a maximum of only 40% loss of IB␣ was observed ( Fig. 2A, middle panel). By 1 h, IB␣ protein levels had returned to above resting levels, which is consistent with NF-B-dependent activation of the IB␣ gene (17,18). In marked contrast, TNF-␣ caused a rapid loss of IB␣ within 15 min of stimulation, and again resynthesis was observed 1 h post-stimulation ( Fig. 2A, right panel). These data agree with the fact that NF-B DNA binding was induced more rapidly by TNF-␣ than by IL-1␤. However, by 30 min, NF-B DNA binding was essentially similar with both stimuli, yet loss of IB␣ by TNF-␣ stimulation was almost total, whereas IL-1␤ resulted in a mere 40% reduction. Furthermore, by 15 min, IL-1␤ resulted in only a 15% loss of IB␣ yet caused over 60% relative DNA binding activity. One explanation for these discrepancies could be that IL-1␤-dependent induction of IB␣ resynthesis occurred more rapidly such that total disappearance of IB␣ was prevented. This possibility was addressed by stimulation in the presence of cycloheximide to prevent new protein synthesis (Fig. 2B). In control cells, cycloheximide had little effect on IB␣ protein levels (Fig. 2B, left panel). After IL-1␤ treatment in the presence of cycloheximide, loss of IB␣ protein was less than 50% that of control at 30 min. This contrasts with the complete loss of IB␣ within 15 min of TNF-␣ treatment in the presence of cycloheximide.
The above data raise the possibility that IB␣-dependent release of NF-B may not fully account for the induction NF-B DNA binding activity observed after IL-1␤ treatment in U937 (1 ng/ml) or TNF-␣ (10 ng/ml) for the times indicated. Cytoplasmic proteins were prepared, and 20 g were subjected to 10% SDS-PAGE. Immunoblotting revealed IB␣ as a 37-kDa protein, and representative blots are shown. After densitometry, data (n ϭ 4) were expressed as a percentage of control as means ϮS.E. and are shown below. B, cells were treated as above except that cycloheximide (10 g/ ml) (CHX) was added 5 min before stimulation. Representative immunoblots for IB␣ are shown, and data (n ϭ 3) are plotted below as in A. Cells were treated as in A, and immunoblotting was performed for IB␤ (n ϭ 4) (C), IB⑀ (n ϭ 4) (D), and p105 (n ϭ 3) (E). In each case, proteins were detected as bands at 42 kDa, 45 kDa, and 105 kDa respectively. Representative immunoblots are shown, and data is plotted below as in A. Open bars, IL-1␤; solid bars, TNF-␣. cells. We therefore focused on the potential role of other IB proteins in activation of NF-B. At 60 min, IB␤ showed a maximal loss of 30 -40% with IL-1␤ and more than 50% loss after TNF-␣ treatment (Fig. 2C). The kinetics of TNF-␣-induced loss of IB␤ were delayed with respect to loss of IB␣ and the repression of IB␤ protein levels, although modest, were more prolonged. The effects of IL-1␤ on loss of IB␤ were less pronounced than for TNF-␣ and only observed around 1-2 h post-stimulation. Consequently, IB␤ does not appear to contribute to levels of activated NF-B observed before 1 h after IL-1␤ treatment.
Again, relatively minor decreases were observed for IB⑀ after IL-1␤ and TNF-␣ treatments, and in each case, the kinetics were similar, suggesting that IB⑀ does not play a major role in release of active NF-B (Fig. 2D). In addition, degradation of the p50 precursor, p105, via the ubiquitin pathway may release active p50/p65 heterodimers (19), whereas in some cells types, including lymphoid cells, the C-terminal part of p105 can be independently transcribed to produce IB␥ (20,21). Western blot analysis with an antibody for IB␥, which also reacts with p105, revealed only p105 in U937 cells. Loss of p105 after treatment with IL-1␤ or TNF-␣ occurred with similar kinetics and in each case was no more than 50% (Fig. 2E). In addition, a p50 antibody that cross-reacts with p105 also detected p105 and gave similar results to the IB␥ antibody (data not shown). Furthermore and consistent with the supershift data, immunoblot analysis failed to detect either p52 or precursor protein p100 (data not shown). These data suggest that IB␤, IB⑀, or p105 may not be the major sources of active NF-B observed on EMSA after IL-1␤ or TNF-␣ stimulation in U937 cells.

IB␣ Remains Bound to p65 and Is Localized in the Cytoplasm after IL-1␤ Stimulation-In
Jurkat T-cells, pervanadate causes release of NF-B without degradation of IB␣, and this event involves tyrosine phosphorylation rather than serine phosphorylation of IB␣ (22). We therefore used immunoprecipitation of p65 followed by Western blotting for IB␣ to investigate whether IL-1␤-stimulated release of NF-B occurred without degradation of IB␣. After 15 min of IL-1␤ stimulation, but not TNF-␣ stimulation, IB␣ was found to co-precipitate with p65 (Fig. 3A). By 90 min post-stimulation, co-precipitation of IB␣ with p65 was observed for both stimuli, consistent with reappearance of newly synthesized IB␣. The specificity of p65 immunoprecipitation was confirmed by competition with specific blocking peptide (Santa Cruz). These data show that after IL-1␤ treatment, IB␣ is still bound to p65, whereas after TNF-␣ treatment, there is no IB␣ bound to p65.
Some studies have suggested that IB␣ may translocate to the nucleus (23). Consequently, we have used Western blot analysis of nuclear extracts to explore the possibility that after IL-1␤ treatment, p65 translocates to the nucleus although still bound to IB␣. In resting cells, p65 was not detected in nuclear extracts. However, nuclear translocation of p65 was readily detected at 15, 30, and 60 min after both IL-1␤ and TNF-␣ treatments (Fig. 3B). In either case nuclear IB␣ was not detected (even after prolonged overexposure of film), indicating that translocation of p65 was not accompanied by IB␣.
Immunoblotting of nuclear and cytoplasmic protein for p65 suggested that the level of cytoplasmic p65 in untreated cells was substantially greater than the nuclear level after either IL-1␤ or TNF-␣ treatment (Fig. 3B). Indeed, immunoblot analysis of cytoplasmic extracts after IL-1␤ or TNF-␣ treatments revealed only modest decreases in p65 immunoreactivity, indicating that only a fraction of the total cytoplasmic p65 was involved in nuclear translocation (Fig. 3C).
Differential Activation of the NIK-IKK Pathway by IL-1␤ and TNF-␣-Various studies have shown that IL-1␤ and TNF-␣ stimulation converge at the level of NIK and cause IKK activation leading to phosphorylation of IB␣ on serines 32 and 36 and its rapid degradation (6 -12, 24). However in U937 cells, rapid agonist-dependent degradation of IB␣ was not observed after IL-1␤ treatment, suggesting that there may be a defect in the signaling pathway leading to IB␣ degradation.
Western blot analysis using an antibody specific for the Ser32-phosphorylated form of IB␣ showed phosphorylated IB␣ at 5 min and at 60 min after TNF-␣ stimulation (Fig. 4A). This is consistent with a rapid IKK-dependent phosphorylation of IB␣ before its degradation and the continued phosphorylation and turnover of newly synthesized IB␣ 60 min after TNF-␣ stimulation (Fig. 4A). The IB␣ species detected by the phospho-Ser32-specific antibody coincides exactly with the reduced mobility band observed with the pan-IB␣ antibody, confirming that this reduced mobility was indeed the result of serine 32 (and presumably 36) phosphorylation. By contrast, the phospho-Ser32-specific antibody failed to detect IB␣ at any time point after IL-1␤ stimulation (Fig. 4A). Furthermore, IB␣ phosphorylation at other sites, for example tyrosine 42 or other C-terminal serine residues (22,25,26), seems unlikely as no evidence of a mobility shift was observed after IL-1␤ treatment. These data strongly suggest that the IB␣ serine 32 and 36 specific IKK kinase activity is active in TNF-␣-treated but not in IL-1␤-treated cells.
To examine this possibility, IKK signalsomes were immunoprecipitated using an anti-IKK␣ antibody and IB␣ kinase

FIG. 3. Effect of IL-1␤ and TNF-␣ on p65/IB␣ binding and cellular localization.
A, immunoprecipitation (IP) of p65 was performed using cytoplasmic extracts (100 g) from cells that were either not stimulated or treated with IL-1␤ (1 ng/ml) or TNF-␣ (10 ng/ml), as indicated. Immunoprecipitates were subjected to Western blot analysis for IB␣ (lower panel) before stripping, and immunodetection was performed for p65 (upper panel). Specificity was confirmed by incorporation of p65 antibody blocking peptide (BP) in the immunoprecipitation. Heavy chain IgG is indicated. B, cells were either not stimulated or treated with IL-1␤ (1 ng/ml) or TNF-␣ (10 ng/ml) for the times indicated. Nuclear extracts were prepared, and 20 g of protein was subjected to 10% SDS-PAGE. After immunoblotting for p65 (top panel), membranes were stripped and reprobed with IB␣ (lower panel). Cytoplasmic protein (20 g) from unstimulated cells was used as positive control for IB␣ (cyt). C, cells were treated as in B, and cytoplasmic extracts, 20 g, were used for p65 immunoblotting. In each case (A, B, and C) blots representative of three such experiments are shown. activity assayed by phosphorylation of the substrate GST-IB␣ (1-54) (Fig. 4B). TNF-␣ stimulation dramatically increased IKK activity between 0 and 10 min with maximal activity at around 2 min. Although this rapid increase in IKK activity was not observed after IL-1␤, a gradual increase in IKK activity was observed, reaching around 30% that of the maximal response obtained with TNF-␣ (Fig. 4B). The presence of both IKK␣ and IKK␤ in immunoprecipitates was confirmed by Western blot analysis (Fig. 4B). The IKK␤ antibody detected a specific band at 87 kDa, whereas the IKK␣ antibody detected a major band at 85 kDa and a minor band at 87 kDa corresponding to IKK␣ and cross-reactivity with IKK␤, respectively. The specificity of immunoprecipitation was confirmed using control rabbit IgG antisera, which failed to immunoprecipitate either IKK␣ or IKK␤ (Fig. 4C). Furthermore, no phosphorylation of IB␣ was observed when the substrate, GST-IB␣ (1-54), was either not present or replaced with the mutant GST-IB␣(1-54; S32A, S36A) (Fig. 4C). The substrate specificity was further confirmed using partially activated IKK signalsome that had been biochemically purified from okadaic acid-stimulated HeLa cells (27). These results demonstrate that the rapid and transient IKK activation, after TNF-␣ treatment in U937, does not occur after IL-1␤ stimulation of U937 cells.
The above result indicates that the signaling pathway leading to activation of IKK may be impaired in response to IL-1␤. As the upstream kinase, NIK, is thought to directly phosphorylate IKKs (11,12), the role of NIK in IL-1␤-and TNF-␣-dependent NF-B activation was examined by co-transfecting U937 cells with a B-dependent reporter and plasmids expressing either wild type or dominant negative versions of NIK (9). Wild type NIK strongly induced B-dependent transcriptional activity, which is consistent with a role for NIK in the NF-Binducing pathway (Fig. 5A). Overexpression of dominant negative NIK inhibited TNF-␣-induced B-dependent transcriptional activity by 60.3 Ϯ 3.2% but only inhibited IL-1␤-induced activity by 33.5 Ϯ 9.2%. (Fig. 5, B). This discrepancy suggests that NIK may be less important in the signaling pathway leading to B-dependent transcription induced by IL-1␤ than by TNF-␣.
Effect of Proteasome, Src, and Tyrosine Kinase Inhibitors on NF-B DNA Binding and Transcriptional Activation-The above results raise the possibility that in U937 cells, IL-1␤-induced NF-B is released from a pool of molecules that is distinct from the pool associated with IB␣. As the ubiquitinproteasome system is not only required for proteolysis of IB␣ but also the proteolysis of p105 and most probably other IB molecules such as IB␤ or ⑀ (19,28,29), the inhibitor, MG-132, was used to test for proteasome involvement in IL-1␤-dependent activation of NF-B in U937 cells. This class of compound are potent inhibitors of the 26 S proteasome, and MG-132 has previously been shown to inhibit activation of NF-B and IL-8 production (30 -32). First, the effect of MG-132 was confirmed by its ability to block IB␣ proteolysis induced by TNF-␣ (data not shown). In addition this compound blocked induction of NF-B DNA binding and B-dependent transcriptional activity by both IL-1␤ and TNF-␣, suggesting that 26 S proteasome activity is necessary for both TNF-␣-and IL-1␤-dependent NF-B activation (Fig. 6, A and B).
Recently, phosphorylation of IB␣ at tyrosine 42 via the nonreceptor tyrosine kinase, c-Src, was shown to mobilize NF-B to the nucleus via a mechanism that does not involve IB␣ degradation (22,33). To further exclude this possibility, the selective Src family inhibitor, PP1 (34), and the proteintyrosine kinase inhibitor, genistein, which inhibits induction of

FIG. 4. Analysis of Ser-32-phosphorylated IB␣ and IKK activity after IL-1␤ and TNF-␣.
A, cells were treated with IL-1␤ (1 ng/ml) or TNF-␣ (10 ng/ml) for the times indicated. Cytoplasmic extracts were prepared, and Western blotting (WB) was performed for the phosphorylated form of IB␣ (IB␣-P) (upper panels). Membranes were stripped and reprobed with pan-IB␣ antibody (IB␣) (lower panels). B, cells were treated as in A, and IKKs were immunoprecipitated from the cytoplasmic extract. Immunoprecipitates were divided in two, one-half was subjected to IKK kinase assay (KA) using GST-IB␣ (1-54) as substrate (upper panel), and the other half was subject to Western blot analysis for IKK␣ (middle panel) and ␤ (lower panel). Densitometry was performed for IKK activity, and data is plotted as fold induction. In each case (A and B) data is representative of three independent experiments. C, immunoprecipitates from cells treated with TNF-␣ for 2 min were subjected to IKK assay (KA) using a either wild type GST-IB␣ (1-54), mutant GST-IB␣-(1-54; S32A, S36A), or no substrate as indicated. Kinase assay on immunoprecipitates using control rabbit preimmune antisera (PI) and biochemically purified IKK complex from HeLa cells treated with okadaic acid (IKK) was also performed as indicated.
NF-B-dependent transcription by lipopolysaccharide in THP-1 cells (35), were tested. Neither PP1 nor genistein had any effect on IL-1␤-or TNF-␣-induced NF-B DNA binding, indicating that these signaling pathways are not involved in the activation of NF-B DNA binding (Fig. 7A). However, both PP1 and genistein inhibited B-dependent transcription by IL-1␤ and TNF-␣ to similar extents, suggesting that tyrosine kinase activity, possibly of a Src family kinase, is necessary for NF-B-dependent transcriptional activity.
Analysis of p65 Phosphorylation-As p65 phosphorylation has been shown to positively module the transcriptional activity of NF-B (36 -38), the effect of TNF-␣ and IL-1␤ on p65 phosphorylation was investigated (Fig. 8). As in these previous studies we found p65 to be present as a phospho protein. However, treatment with either IL-1␤ or TNF-␣ for up to 1 h failed to produce any substantial changes in p65 phosphorylation, suggesting that differences in the regulation of NF-B activity are not mediated at this level. DISCUSSION NF-B is widely accepted as playing a key role in inflammation via involvement in both effector and target gene regulation. Consequently, the NIK-IKK-IB␣ signaling pathway now represents a prime candidate for therapeutic intervention (1,2). However, numerous other signaling pathways including protein kinase C, phosphatidylcholine-specific phospholipase, sphingomyelinase, protein kinase A, tyrosine kinases, and mitogen-activated protein kinases are all implicated in the regulation of NF-B (22, 33, 35, 37, 39 -43). Clearly, numerous unresolved questions remain in respect of the events leading to NF-B-dependent transcription. We report that in U937 monocytic cells, TNF-␣ and IL-1␤, both, produced identical B-dependent transcriptional responses, yet the magnitudes and kinetics of IKK activation and IB␣ degradation were totally different. This result essentially leads to two possibilities.
First, IKK activation and IB␣ degradation are not exclusively required for NF-B activation or second, some other step (or steps) that is required for NF-B transcriptional activation is in fact rate-limiting.
In HeLa cells, IKK activation by TNF-␣ and IL-1␤ occurs very rapidly and is compatible with the TNF-␣ response in U937 cells (8). In contrast, IKK activation by IL-1␤ in U937 cells was substantially slower and only reached a peak of less than 30% that of the TNF-␣-induced level. This resembled the effect observed in THP-1 cells after LPS stimulation in which IKK activity was not maximal until 1 h post-stimulation (44). We also examined IKK activity at up to 1 h post-IL-1␤ or TNF-␣ treatment, but no further increases in activity were observed (data not shown). In the above experiments the extent and kinetics of IKK activation and IB␣ degradation correlated very closely. Despite this, no phosphorylated IB␣ was detected in IL-1␤-treated cells, suggesting that this level of IKK activation may not be physiologically relevant and that the loss of IB␣ was via some other mechanism. However, after IL-1␤ treatment in the presence of the proteasome inhibitor MG-132, phosphorylated IB␣ was detected by mobility shift and using a serine 32 phospho-specific antibody (data not shown). Thus, IL-1␤ causes phosphorylation of IB␣, and this is consistent with the low level of IKK activity. Reasons for the difference between the TNF-␣ and the IL-1␤ responses are presently unclear, as in other cell types these cytokines both result in rapid degradation of IB␣ and NF-B activation (7,40,45). However, these reports along with our data and that of O'Connell et al. (44) highlight the existence of two kinetically, and presumably mechanistically, distinct pathways of IKK activation.
In view of the dramatically different kinetics and extents of IKK activation and IB␣ degradation induced by IL-1␤ and TNF-␣, we were surprised to find that TNF-␣ only resulted in around 30 -40% more peak DNA binding activity than IL-1␤. This and other discrepancies showed that loss of IB␣ does not necessarily equate with NF-B DNA binding. It was therefore possible that other IB proteins may also take part in activation of NF-B. The kinetics of IB␤ degradation were consistent with the peak of TNF-␣-induced NF-B DNA binding activity observed around 1-2 h and with previous reports implicating IB␤ in the late or delayed NF-B response (46,47). With IL-1␤ there was little change in IB␤ levels by 30 min, suggesting that this protein does not contribute appreciably to DNA binding observed at 30 min. A number of reports have shown that serines 19 and 23 in IB␤ are substrates for the IKK complex, and their phosphorylation is required for proteolytic degradation (7,27,28). However, IB␤ is a considerably poorer IKK substrate compared with the equivalent IB␣ reaction (7,44). Despite this, it is difficult to explain the fact that, unlike IB␣, the kinetics of IB␤ degradation do not coincide with or follow closely behind the kinetics of IKK activation. Instead there was a considerable delay, suggesting that either phosphorylation of IB␤ by IKK occurs rapidly and concordantly with IKK activity but that the subsequent degradation occurs by a less rapid mechanism or that a IB␤ kinase exists that shows slower activation kinetics compared with the IB␣ kinases. Likewise degradation of both IB⑀ and p105 was limited and similar for both treatments, whereas IB␥ was not detected in these cells. Consequently, these IBs also fail to account for the substantial increase in DNA binding induced by IL-1␤.
The observation that tyrosine phosphorylation of IB␣, possibly mediated by c-Src, may cause release of NF-B from IB␣ without proteolytic degradation could explain the above results (22,33). However, our data exclude involvement of tyrosine phosphorylation and degradation-independent release of NF-B from IB␣ in the induction of NF-B DNA binding in U937 cells. Likewise, possible nuclear translocation of NF-B, although bound to IB␣, was also excluded. Additionally, Western analysis revealed that only a fraction of the total p65 was mobilized to the nucleus on stimulation. Thus, complete loss of IB␣ corresponded with a partial reduction in cytoplasmic p65, indicating that additional mechanisms must exist for retaining p65/NF-B in the cytoplasm. Clearly IB␤, p105, and IB⑀ may contribute to this effect. However, as these IB proteins along with IB␣ do not appear to fully account for the induction of NF-B DNA binding, there could be a role for additional IB genes, such as IBR and IBl (48,49). In this respect, the findings of Baeuerle and Baltimore appear to be particularly salient (50). These authors describe a protein of 60 -70 kDa in pre-B cells, which is the predominant IB activity in these cells. As IB␥ (70 kDa) generally seems to be lowly expressed, if at all, and this protein is unlikely to correspond to IB␣ (37 kDa), IB␤ (45 kDa), or IB⑀ (45 kDa), it may represent an as yet uncloned IB molecule and could account for the effects observed here. Based on the data presented, we therefore predict that release of NF-B from a number of cytoplasmic pools or stores is necessary to produce the overall response. These data are consistent with a role for IKK activation and IB␣ in the immediate TNF-␣-dependent NF-B response and IB␤ in the delayed response. However, there appears to a role for additional IB activities and activating pathways in mediating the intermediate NF-B response. In this case it seems that proteasome activity is required for NF-B activity, as the MG-132 totally prevented IL-1␤-and TNF-␣-dependent NF-B DNA binding. Candidate molecules for this effect may include p105, IB⑀, the putative IBR and IBl genes, and the IB protein identified by Baeuerle and Baltimore (50).
The difference between IB␣ degradation induced by IL-1␤ and TNF-␣ suggested a major difference in the signaling events leading to IB␣ degradation. The greater inhibitory effect of dominant negative NIK on TNF-␣-dependent reporter activity compared with the IL-1␤ effect implies a greater role for NIK in the TNF-␣ response. As NIK is thought to directly activate IKK (10 -12), this hypothesis is in accord with the IKK activity data presented. However, despite testing a range of expression vector concentrations, dominant negative NIK failed to completely block reporter activation by either cytokine, implying that other signaling pathways that are not blocked by dominant negative NIK may also exist. Similar data are also found in respect of both IKK␣ and ␤, as dominant negative versions of these kinases often fail to repress or only partially repress B-dependent responses (7,44). Although these data may be explained by phenomena associated with overexpression, the existence of alternative NF-B-activating pathways also remains a possibility. Finally, this study also highlights one further area of discrepancy. This involves the lack of correlation between NF-B DNA binding and B-dependent transcription. In this study, IL-1␤ produced lower levels of DNA binding than TNF-␣. However, the transcriptional responses to these cytokines were indistinguishable. One possibility is that TNF-␣ and IL-1␤ differentially modulate the transactivation potential of p65 via different degrees of p65 phosphorylation (36,37,51). However analysis of p65 phosphorylation showed no differences between TNF-␣ and IL-1␤ treatments, suggesting that downstream steps may be rate-limiting. This possibility is supported by the fact in monocytes that genistein had no effect on DNA binding yet prevented NF-B transcriptional activation while enhancing AP-1-dependent transcription (35). Likewise in U937 cells, genistein and PP1 both showed little effect on NF-B DNA binding but markedly attenuated transcriptional activation. Furthermore, we and others have previously reported similar effects in other cell types indicating that the control of transcriptional activation per se remains to be explored (35,40).
In conclusion, we have presented a body of data that identifies a number of inconsistencies in the NF-B transcriptional activation cascade. We therefore hypothesize the existence of parallel activation pathways that are distinct from the classical NIK-IKK-IB␣ activation pathway. Indeed, because the completion of this study, a number of reports have shown that in addition to NIK and MEKK1, the mitogen-activated protein kinase kinase kinases, TPL-2, TAK1, MEKK2, and MEKK3, are also able to activate NF-B, further raising the possibility of parallel activation cascades (52)(53)(54). In addition, our data also point to the existence of signaling pathways, which may impose rate-limiting steps on B-dependent transcription. These findings have a number of important implications for anti-inflammatory strategies aimed at inhibiting the NIK-IKK-IB␣ pathway. Drugs that cause relatively high degrees of inhibition of IKK or NIK may in fact only produce relatively modest changes in NF-B activation at the DNA binding level. These changes may be further atten- FIG. 7. Effect of the Src inhibitor, PP1, and the tyrosine kinase inhibitor, genistein, on NF-B DNA binding and B-dependent transcription. A, cells were treated with IL-1␤ (1 ng/ml) or TNF-␣ (10 ng/ml) for 60 min as indicated. PP1 (10 M) or genistein (Gen) (100 M) was added 30 min before stimulation. Nuclear extracts were prepared, and EMSA was performed for NF-B. Specificity of DNA binding complexes, indicated with a solid arrow, was determined by competition with a 100-fold excess of unlabeled probe (100ϫ). An autoradiograph representative of three such experiments is shown. B, cells transfected with pGL3.6B.BG.luc cells were treated as indicated with IL-1␤ or TNF-␣ after pretreatment with PP1, genistein, or vehicle (Me 2 SO) as in A. After 6 h, luciferase assays and protein determination was performed. Data (n ϭ 5) in duplicate are expressed relative to control and plotted as means ϮS.E.
FIG. 8. Effect of TNF-␣ and IL-1␤ on p65 phosphorylation. Cells were incubated in phosphate-free medium for 2 h before addition of [ 32 P]orthophosphate at 0. 2 mCi/ml for a further 4 h. After stimulation with IL-1␤ (1 ng/ml) and TNF-␣ (10 ng/ml) for the times indicated, immunoprecipitation of p65 was performed, and samples were divided into two and analyzed by SDS-PAGE followed by autoradiography (AR) or Western blotting for p65 (WB). The autoradiograph and corresponding blot is representative of three similar experiments. uated by downstream rate-limiting steps that are involved in transcriptional activation itself. The generation of highly selective NIK and IKK inhibitors along with the continued cloning of components of the NF-B activation cascade will be required to answer these questions.