Bcl-3 and NFκB p50-p50 Homodimers Act as Transcriptional Repressors in Tolerant CD4+ T Cells*

The transcriptional events that control T cell tolerance are still poorly understood. To investigate why tolerant T cells fail to produce interleukin (IL)-2, we analyzed the regulation of NFκB-mediated transcription in CD4+ T cells after tolerance induction in vivo. We demonstrate that a predominance of p50-p50 homodimers binding to the IL-2 promoter κB site in tolerant T cells correlated with repression of NFκB-driven transcription. Impaired translocation of the p65 subunit in tolerant T cells was a result from reduced activation of IκB kinase and poor phosphorylation and degradation of cytosolic IκBs. Moreover, tolerant T cells expressed high amounts of the p50 protein. However, the increased expression of p50 could not be explained by activation-induced de novo synthesis of the precursor p105, which was constitutively expressed in tolerant T cells. We also demonstrate the exclusive induction of the IκB protein B cell lymphoma 3 (Bcl-3) in tolerant T cells as well as its specific binding to the NFκB site. These results suggest that the cellular ratio of NFκB dimers, and thus the repression of NFκB activity and IL-2 production, are regulated at several levels in tolerant CD4+ T cells in vivo.

The transcriptional events that control T cell tolerance are still poorly understood. To investigate why tolerant T cells fail to produce interleukin (IL)-2, we analyzed the regulation of NFB-mediated transcription in CD4 ؉ T cells after tolerance induction in vivo. We demonstrate that a predominance of p50-p50 homodimers binding to the IL-2 promoter B site in toler-

ant T cells correlated with repression of NFB-driven transcription. Impaired translocation of the p65 subunit in tolerant T cells was a result from reduced activation of IB kinase and poor phosphorylation and degradation of cytosolic IBs. Moreover, tolerant T cells expressed high amounts of the p50 protein. However, the increased expression of p50 could not be explained by activation-induced de novo synthesis of the precursor p105, which was constitutively expressed in tolerant T cells. We also demonstrate the exclusive induction of the IB protein B cell lymphoma 3 (Bcl-3) in tolerant T cells as well as its specific binding to the NFB site. These results suggest that the cellular ratio of NFB dimers, and thus the repression of NFB activity and IL-2 production, are regulated at several levels in tolerant CD4 ؉ T cells in vivo.
Induction of T cell tolerance may result in either deletion or anergy, the latter being characterized by a block in proliferation and diminished expression of Th1 cytokines, particularly interleukin-2 (IL-2) 1 (1). Although T cell anergy has been described both in vitro and in vivo (2)(3)(4)(5), recent studies have shown that rather than being functionally inert, the anergic T cells may adopt the role of regulatory T cells with a Tr1-like phenotype (6,7). Importantly, anergic T cells may be functionally regulated at several different levels, e.g. in vivo tolerized T cells may neither produce IL-2 nor respond to signaling through the IL-2 receptor (8).
Because the failure of tolerant T cells to produce IL-2 is a hallmark of anergy, it is particularly important to delineate the molecular explanation for this defect. Downstream of the T cell receptor (TCR), several intracellular signaling pathways converge at the level of transcriptional regulation of the IL-2 gene. AP-1, NFAT, and NFB families of transcription factors are essential in T cell activation as coordinators of IL-2 transcription (9) and tolerant CD4 ϩ T cell clones as well as in vivo tolerized T cells possess reduced AP-1 binding to the IL-2 promoter (10 -12). In addition, we have previously shown that CD4 ϩ T cells from mice tolerized in vivo express qualitative alterations in the composition of the NFB complex (13). Defective NFB expression is also a feature of tolerant B cells (14,15).
The NFB family consists of five members, p65, c-rel, RelB, p105/p50, and p100/p52, and the NFB complex is formed by hetero-or homodimerization of these different proteins (16). p65, c-rel, and RelB are synthesized as mature proteins and contain transactivation domains, whereas p105/p50 and p100/ p52 are produced as inactive precursor proteins and lack transactivation domains. Activated NFB is generally composed of p65-p50 or c-rel-p50 heterodimers, whereas homodimeric complexes of p50-p50 or p52-p52 are associated with transcriptional repression (17)(18)(19). In this respect, p50-p50 homodimers have been implicated both in LPS-induced tolerance in vitro (20,21) and in superantigen-induced T cell tolerance in vivo (13). In resting T cells, the active NFB dimers are bound to the IB inhibitory proteins, e.g. IB␣ and IB␤, and hence retained in the cytosol. Upon cellular stimulation, the IB proteins are rapidly phosphorylated and targeted for degradation by the proteasome, thereby releasing NFB, which then translocates into the nucleus and activates transcription (16,22). The stimulus-induced phosphorylation of the IBs is mediated by a 700-kDa complex called the IKK, which consists of three major components, the catalytic subunits IKK␣ and ␤ and a regulatory subunit IKK␥ (NEMO) (22,23). NFB activation may also be regulated by the IB protein Bcl-3, a putative proto-oncogene (24). Unlike IB␣ and -␤, which associate with active NFB dimers in the cytosol, Bcl-3 specifically associates with transcriptionally inactive p50-p50 or p52-p52 homodimers in the nucleus (25)(26)(27). However, the precise role of Bcl-3 is controversial, with some reports suggesting that binding of Bcl-3 to p50-p50 homodimers promotes transcription (27,28), whereas other studies indicate that Bcl-3 has a negative effect on NFB-dependent transcription (29,30).
Here we demonstrate that lack of NFB-dependent transcription in tolerant CD4 ϩ T cells in vivo correlated with defective expression of IL-2. Activation of IKK, degradation of IBs, and nuclear translocation of p65 were inhibited in tolerant T cells, which expressed higher levels of the p50 protein both in the cytosol and nucleus compared with activated T cells. Because transcriptional induction of the precursor protein p105 did not differ between tolerant and activated T cells, our data suggest that the overall increased levels of p50 in tolerant T cells may be due to post-translational modifications of the p105 protein. Moreover, Bcl-3 was induced both at the mRNA and protein levels in tolerant T cells, and its specific binding to the B site proposes a regulatory function of Bcl-3 for p50-p50 homodimers in tolerant T cells.

EXPERIMENTAL PROCEDURES
Animals and Treatment-TCR V␤3 transgenic mice expressing a rearranged genomic clone of the 2B4 ␤-chain gene under the influence of an inserted Ig heavy chain enhancer (31) were generously provided by Dr. M. Davis (Stanford). The CD4 ϩ T cell population was typically Ͼ95% TCR-V␤3 ϩ . Dr. R. Flavell (Howard Hughes Medical Institute, Yale) generously provided the NFB-luciferase reporter transgenic mice (NFB-luc). These mice were generated using the pBIIX-luciferase construct with two copies of the B sequences from the Ig enhancer (32). The NFB-luc mice were backcrossed with TCR-V␤3 transgenic mice, and double-transgenic mice were used for experiments. Recombinant SEA was produced as described previously (33). Ten g of SEA in 0.2 ml of phosphate-buffered salin with 1% normal syngeneic serum were injected intravenously (i.v.) at 4-day intervals. SEA was injected three times to induce tolerance and once to induce T cell activation.
Cell Separation-For purification of CD4 ϩ T cells, cell suspensions of spleen and lymphnodes were prepared from mice injected i.v. with SEA at different times prior to analysis. CD11c ϩ dendritic cells (DCs) were isolated from single-cell suspensions of spleens pretreated with collagenase (Chemicon) and 1% DNase (Sigma) for 30 min at 37°. Purified CD4 ϩ T cells (Ͼ95% CD4 ϩ as determined by fluorescence-activated cell sorter analysis) and CD11c ϩ dendritic cells (Ͼ90% CD11c ϩ ) were obtained by positive selection using magnetic beads coated with anti-CD4 mAb or anti-CD11c mAb (Miltenyi Biotech GmbH, Bergisch Gladbach, Germany).
ELISA-IL-2 in culture supernatants and serum was determined by ELISA (BD Biosciences). p65 and p50 EMSA ELISAs were performed using a TransAM NFB kit from Active Motif (Carlsbad, CA). To prevent unspecific binding of the polyclonal p50 and Bcl-3 mAbs used in this assay, mAbs were preincubated with poly(dI-dC), herring sperm DNA, and mutant B oligonucleotide before addition to the ELISA plate.
Luciferase Activity Analysis-Purified CD4 ϩ T cells from NFB-luc mice were plated into flat-bottomed 96-well tissue culture plates using 5 ϫ 10 5 cells/0.2 ml of R10 medium lacking phenol red. The CD4 ϩ T cells were cultured in the presence or absence of SEA (1 g/ml) or PMA (5 ng/ml) and ionomycin (250 ng/ml). Purified CD11c ϩ dendritic cells from NFB-luc-negative mice were used as antigen presenting cells at 0.5 ϫ 10 5 /well. In the inhibitor experiments, CD4 ϩ T cells with indicated concentrations of inhibitors were preincubated for 30 min-1 h at 37°C before addition of SEA-pulsed DCs. After 16 -24 h, cells were washed twice in phosphate-buffered saline and then lysed in lysis buffer (luciferase assay; Promega, Madison, WI) for 5-15 min at 4°C. The lysates were analyzed using the luciferase reagent (Promega) and measured in a luminometer (LUMIstar Galaxy; bmg, LabVision). Experimental values are expressed as recorded light units of luciferase activity.
Quantitative Transcript Analysis-Total RNA was extracted from purified CD4 ϩ T-cells using TRIZOL (Invitrogen) according to the manufacturer's instructions and was reverse-transcribed using random hexamer primers (Amersham Biosciences). Quantitative real-time PCR was performed using a ROCHE LightCycler (Roche Diagnostics) utilizing SYBR Green 1 (Roche Diagnostics) for the sequence-independent detection of DNA according to the manufacturer's instructions. In brief, reactions (20 l) containing 2 l of cDNA, 3 mM MgCl 2 , 150 g/ml bovine serum albumin (Roche Diagnostics), 1 unit of Platinum TaqDNA polymerase (Invitrogen) were denatured at 95°C for 5 min, followed by 35 cycles of amplification. Primer annealing temperatures ranged between 60 -65°C.
Preparation of Cellular, Cytosolic, and Nuclear Extracts-Whole-cell extracts as well as cytosolic and nuclear fractions used for immunoprecipitations, Western blots, EMSA, and ELISA were made from 10 -15 ϫ 10 6 purified CD4 ϩ T cells. Protein concentrations in different groups within the same experiment were normalized by equal cell numbers and/or measuring protein content by the Bio-Rad protein assay kit (Bio-Rad Laboratories). CD4 ϩ T cells were purified after indicated periods of time ex vivo. In the case of in vitro restimulation, CD4 ϩ T cells were incubated with 10% DCs and SEA (1 g/ml) for different time points in a 37°C-water bath. Whole-cell extracts and nuclear extracts were prepared as previously described (8,13). As an alternative preparation of cytosolic and nuclear extracts, a nuclear extraction kit from Active Motif was used according to the instructions of the manufacturer.
Immunoprecipitation and Western Blotting-Cell lysates were precleared for 30 min using 20 -30 l/sample of protein A-Sepharose CL-4B beads (Pharmacia Biotech). The precleared supernatants were incubated with the indicated antibody for 1 h rotating at 4°C. Protein A beads were added, and the samples were incubated for an additional 1-2 h. The immunoprecipitated proteins were pelleted and washed three times in lysis buffer. Finally, the proteins were extracted by the addition of 2ϫ SDS sample buffer and incubation for 10 min at 70°C. Immunoprecipitated, cytosolic, or nuclear samples were subjected to analysis by SDS-PAGE using 4 -12% Bis-Tris gels (Invitrogen). The separated proteins were electroblotted onto nitrocellulose membranes (0.2 M; Bio-Rad) and visualized as described elsewhere (8). For immunoprecipitation of NFB-binding proteins in nuclear extracts, an agarose-conjugated consensus B-oligonucleotide (GGG GAC TTT CCC) was used according to instructions from the manufacturer (Santa Cruz Biotechnology).
Immunocomplex Protein Kinase Assay-Cellular extracts were immunoprecipitated with polyclonal antibody to IKK␣ or IKK␣/␤ as described above. Immune complexes were washed twice in ice-cold lysis buffer and then twice in ice-cold kinase buffer (20 mM HEPES, pH 7.6, 50 mM NaCl, 6 mM MnCl 2 , 5 mM MgCl 2 , 1 mM dithiothreitol, 1 mM EDTA, 1 mM p-nitrophenylphosphate, 10 mM ␤-glycerophosphate, and 0.1 mM Na 3 VO 4 .). The beads were pelleted and resuspended in 15 l of kinase buffer containing 1 M ATP, 1 Ci of [␥-32 P]ATP and 1 g of GST-IB␣. Incubations were carried out for 30 min at 30°C; reactions were stopped by addition of 4ϫ SDS sample buffer. IB␣ phosphorylation was analyzed by 4 -12% SDS-PAGE followed by autoradiography.
Electrophoretic Mobility Shift Assay-The B binding site oligonucleotide used was derived from the murine IL-2 promoter and contained the following sequence: 5Ј-ACCAAGAGGGATTTCACCTAAATCC-3Ј. The EMSA and supershift were performed as previously described (8) and exposed to autoradiography at Ϫ70°C.

Tolerant CD4 ϩ T Cells Express Altered NFB Binding to the B Site of the IL-2
Promoter-In this study, we used a well established model of superantigen-induced T cell tolerance (8,12,13,34,35) to investigate downstream defects in TCRmediated regulation of the IL-2-gene in tolerant CD4 ϩ T cells. It has previously been shown that IL-2 protein expression is strongly reduced at all times in tolerant (3ϫ SEA) compared with activated (1ϫ SEA) T cells, both in culture supernatants and in serum samples (8, 13) (data not shown). Furthermore, we reported that although activated CD4 ϩ T cells expressed primarily p65-p50 heterodimers, anergic CD4 ϩ T cells expressed high levels of inactive p50-p50 homodimers (13). To extend this finding, we performed gel shift and supershift assays using an oligonucleotide specific for the B site of the murine IL-2 promoter (36). The kinetics of NFB expression revealed that the induction of p50-p50 homodimers was clearly stronger in in vivo tolerized compared with activated T cells, which contained mainly p65-p50 heterodimers (Fig. 1A). Be-cause the 0-h time point is represented by T cells from 2ϫ SEA-treated mice in the 3ϫ SEA-treated group and by T cells from untreated mice in the 1ϫ SEA-treated group, it was intriguing to find a difference in p50-p50 homodimer expression in these resting cells (Fig. 1A). Supershift analysis confirmed that the slower migrating DNA-protein complex consisted of both p65 and p50 and no c-rel, whereas the faster migrating band was made of p50 (Fig. 1B). In addition to p65-p50 heterodimers, the slower migrating band may contain p65-p65 homodimers (Fig. 1B). The lack of c-rel was confirmed by Western blotting, demonstrating no difference in c-rel ex-pression in the cytoplasm of tolerant and activated T cells and the absence of c-rel in the nucleus of either T cell subset (data not shown). Control IgG had no effect on the complex, confirming the specificity of antibody binding (Fig. 1B). Collectively, these findings demonstrate a qualitative difference in NFB complexes binding to the IL-2 promoter in tolerant and activated T cells.
The NFB Transcriptional Activity Is Severely Impaired in Tolerant CD4 ϩ T Cells-In the next set of experiments, we determined the actual transcriptional activity of NFB by using double-transgenic TCR-V␤3xNFB-luc mice, expressing a luciferase reporter gene under the influence of NFB binding sites. Twenty-four hours after the last SEA injection, the CD4 ϩ T cells were isolated and restimulated in vitro in the presence of SEA and NFB-luc negative DCs and analyzed for NFBdriven transcriptional activity. Evidently, activated T cells displayed considerable transcriptional activity in response to SEA, which also correlated with production of IL-2 in the same cultures (Fig. 2, A and B). In contrast, the NFB-driven transcriptional activity in tolerant T cells was greatly inhibited ( Fig. 2A). IL-2 measured from the same cultures was completely suppressed, which reinforces the connection between NFB transcription and IL-2 production (Fig. 2B). However, the transcriptional activity as well as IL-2 production of tolerant T cells was restored upon combined treatment with PMA and ionomycin (Fig. 2, A and B). These results suggest that both a direct activation of PKC/Ras and a rise in intracellular calcium are required to circumvent the transcriptional defect of NFB in tolerant T cells. To confirm that the inhibition of NFB-driven transcriptional activity was specific, the effects of NFB inhibitors on SEA-induced transcription were investigated. The mice were injected once with SEA to induce T cell activation; 24 h later, CD4 ϩ T cells were put into culture as described above in the presence or absence of the indicated inhibitors. As expected, SN50, an inhibitor peptide that blocks the entrance of p50 into the nucleus, inhibited NFB activity, whereas the control peptide (SN50M) had no effect (Fig. 2C). This clearly demonstrates the importance of the p50 subunit in the activated NFB complex. Additionally, inhibiting the association of the regulatory IKK␥-subunit NEMO with IKK␣/␤ by a membrane-permeable blocking peptide also suppressed transcription (Fig. 2C). None of the tested inhibitors affected the viability of cells (data not shown). Importantly, the effects of the NFB inhibitors correlated well with IL-2 production, highlighting again the connection between NFB transcriptional activity and the levels of IL-2 protein products.
Differential Expression of p65 and p50 in the Nucleus of Tolerant and Activated CD4 ϩ T Cells-To investigate whether the defective transcriptional activity of NFB could be due to differential expression of p65 and p50, we examined the cytosolic and nuclear levels of p65 and p50 protein. Western blot analysis revealed a significant decrease of p65 protein in the cytosol upon stimulation of activated T cells, whereas this decrease was less pronounced in the tolerant T cells (Fig. 3A,  left panel). In contrast, the levels of nuclear p65 were higher in activated compared with tolerant T cells (Fig. 3A, right panel), suggesting a more efficient release and translocation of cytosolic p65 to the nucleus of activated T cells. An overall decrease in cytosolic p50 was observed in both tolerant and activated T cells after SEA treatment (Fig. 3A, left panel). However, although the amount of p50 protein in tolerant T cells was rather constant, nuclear p50 slowly increased after SEA treatment in activated T cells (Fig. 3A, right panel). Furthermore, immunoprecipitation of nuclear extracts with a specific consensus B oligonucleotide revealed that considerably more p65 was able to bind to the B motif in activated T cells, whereas p50 was the FIG. 1. Qualitative differences in NFB binding to the IL-2 promoter in tolerant versus activated CD4 ؉ T cells. TCR V␤3 transgenic mice were treated one or three times with 10 g of SEA at 4-day intervals. Spleens and lymph nodes were removed at different time points after the last SEA injection, and nuclear extracts were prepared from purified CD4 ϩ T cells. A, gel shift analysis was performed using nuclear extracts from the indicated periods of time and a 32 P-labeled IL-2 B oligonucleotide. B, supershift analysis was performed using nuclear extracts prepared 1 h after the last SEA injection and 32 P-labeled IL-2 B oligonucleotide in the presence or absence of supershift antibodies against p50, p65, c-rel, and Fra-1 (control IgG). Arrows indicate positions of specific DNA-protein complexes. One of three similar experiments is shown. main factor detected binding to the B motif in tolerant T cells (Fig. 3B). To further study the complex formation of NFB in the cytosol of SEA-treated T cells, cytosolic extracts were immunoprecipitated with p50 and blotted with p65 or IB␣ mAbs. Significantly less p65 as well as IB␣ was associated with p50 in the cytosol of activated compared with tolerant T cells, supporting the assumption of an intact release and translocation of active NFB upon stimulation (Fig. 3C).
As a more sensitive and quantitative method of protein detection, we used specific p65 and p50 EMSA ELISAs. This method analyses the ability of NFB to bind to consensus B oligos pre-bound to the bottom of ELISA plates. In correlation with the Western blot results, the levels of p65 were indeed reduced in the cytosol while increasing in the nucleus of activated T cells (Fig. 4A, upper tables). The overall amount of p50 was considerably higher in tolerant T cells, even in the resting state, implying that de novo synthesis of p105/p50 had already taken place in these cells (Fig. 4A, lower tables). The specificity of B binding was confirmed by incubation with excess mutated or consensus B oligonucleotide (Fig. 4B). Taken together, the above results demonstrate that although the activation-induced nuclear translocation of p65 is considerably more efficient in activated T cells, the cellular levels of p50 are higher in tolerant T cells.
Poor Degradation of IB␣ and IB␤ in Tolerant CD4 ϩ T Cells-According to the Western blots of immunoprecipitated p50, a complex containing p50, p65, and IB␣ was present in the cytosol of tolerant T cells even after stimulation (Fig. 3C),

FIG. 2. Reduced NFB-driven luciferase activity and IL-2 production in tolerant CD4 ؉ T cells.
NFB-lucxV␤3 transgenic mice were injected one or three times i.v. with 10 g of SEA at 4-day intervals. Spleens and lymph nodes were removed 24 h after the last SEA injection, and purified CD4 ϩ T cells were prepared for in vitro cultures. T cells (5 ϫ 10 5 /well) were cultured in medium or stimulated with SEA (1 g/ml), PMA (5 ng/ml), ionomycin (250 ng/ml), or the combination for 16 -24 h of culture. Alternatively, the cells were incubated with different NFB inhibitors: SN50 (inhibits nuclear translocation of NFB), SN50 M (inactive control of SN50), or NEMO (inhibitor of IKK␥ association with IKK␣/␤). Purified V␤3-luc Ϫ DCs (5 ϫ 10 4 /well) were used as antigen presenting cells. A, cells were lysed, and luciferase activity was measured. Experimental values are expressed as recorded light units. B, IL-2 protein levels in culture supernatants were measured by ELISA. C, transcriptional activity (recorded light units) and IL-2 production (units/ml) were measured in parallel cultures after incubation with different inhibitors or control. S.D. were less than 10% of the mean. One of three similar experiments is shown.
FIG. 3. Cellular dynamics of p65 and p50 measured by Western blot. TCR V␤3 transgenic mice were injected one or three times i.v. with 10 g of SEA at 4-day intervals. Spleens and lymph nodes were removed at the indicated periods of time after the last SEA injection. CD4 ϩ T cells were isolated from single-cell suspensions, and cytosolic and nuclear extracts were prepared. A, immunoblotting of cytosolic and nuclear extracts with antibodies against p65 or p50. B, nuclear extracts from SEA-treated mice were immunoprecipitated with a B-specific oligonucleotide for 1 h at room temperature. Precipitated proteins were eluted with 2ϫ SDS sample buffer, and immunoblotted proteins were detected with antibodies against p65 or p50. C, cytosolic extracts from SEA-treated mice were immunoprecipitated with a p50-specific antibody. Precipitated proteins were eluted with 2ϫ SDS sample buffer, and immunoblotted proteins were detected with antibodies against p65 or IB␣. In parallel, cytosolic samples were immunoblotted with a ␤-actin-specific mAb. One of four similar experiments is shown.
implying a perturbed release of transcriptionally active NFB in these cells. To further examine a possible upstream defect in the NFB pathway, we analyzed the cytosolic expression of IB␣ and -␤. In activated T cells, IB␣ degradation was already induced 15 min after SEA administration, and at this time point IB␣ was also partially degraded in tolerant T cells (Fig.  5A, left panel). Later on, the IB␣ protein returned to background levels, consistent with the proposed feedback loop of NFB-driven synthesis of IB␣ (16). The degradation of IB␤ in activated T cells was less pronounced compared with IB␣ and peaked later (Fig. 5A, middle panel). This is probably because IB␣ is more important during the transient activation of NFB, whereas IB␤ is responsible for prolonged NFB activation. Interestingly, IB␤ was almost completely resistant to degradation in the tolerant T cells. Furthermore, a stronger phosphorylation of the IB␣ protein was found in activated compared with tolerant T cells (Fig. 5B). The inhibition of IB phosphorylation and degradation in tolerant T cells provides an explanation for the hampered translocation of active p65-p50 NFB complexes in these cells.
In Vivo Tolerized CD4 ϩ T Cells Fail to Activate the IKK Complex-The IB proteins are phosphorylated by the specific IB kinases, IKK␣ and -␤. Analysis of IKK activity already revealed a strong activation-induced response in activated CD4 ϩ T cells after 15 min of stimulation (Fig. 6). The activity of IKK was evident both at the level of GST-IB␣ phosphorylation (Fig. 6, upper panel) and as autophosphorylation of the IKK complex (Fig. 6, lower panel). In contrast, tolerant T cells failed to activate the IKK complex (Fig. 6). No differences in IKK␣ and -␤ protein levels were detected (Fig. 6, lower panel). These findings suggest that reduced activation of IKK is responsible for the poor phosphorylation and degradation of the IB proteins in tolerant T cells.
In Vivo Induction of Bcl-3 in Tolerant CD4 ϩ T Cells-The IB protein p105 has a dual function in mediating cytoplasmic retention of p50 as well as serving as the precursor molecule of p50. Thus, the increased expression of p50 in tolerant CD4 ϩ T cells could reflect de novo synthesis of p50 from p105 following tolerance induction. In addition, p50-p50 homodimers could associate specifically with the nuclear IB protein Bcl-3 in order to regulate their transcriptional activity. To evaluate the in vivo relevance of these IB proteins, real-time PCR was performed. p105 mRNA was induced in both tolerant and activated T cells after SEA stimulation in vivo (Fig. 7A, upper  table). In contrast, there was a specific induction of Bcl-3 only in tolerant T cells (Fig. 7A, middle table). As expected, IL-2 mRNA was lower in tolerant T cells (Fig. 7A, lower table). Morevover, immunoblotting revealed a constant expression of p105 protein in tolerant T cells, whereas p105 was reduced in activated T cells upon stimulation (Fig. 7B, left panel). This compares with the expression of p50 protein, which was constant in tolerant while increasing in activated T cells after SEA treatment (Fig. 7B, middle panel). However, the initial amount of p50 in resting T cells was considerably higher in the tolerant group, suggesting that the de novo production of p50 possibly had already taken place (Fig. 7B, middle panel). Thus, p105 was expressed in both tolerant and activated T cells but only actively degraded in the latter subset. In accordance with the mRNA expression, EMSA ELISA revealed that the Bcl-3 protein was induced after stimulation of tolerant but not activated T cells (Fig. 7C). The specificity of the ELISA was verified as described (Fig. 7D). Thus, the significant induction of Bcl-3 in tolerant T cells and its specific binding to the B site imply that Bcl-3 is interacting with p50-p50 homodimers and possibly FIG. 4. Cellular dynamics of p65 and p50 measured by specific ELISAs. TCR V␤3 transgenic mice were injected one or three times i.v. with 10 g of SEA at 4-day intervals. Spleens and lymph nodes were removed at the indicated periods of time after the last SEA injection. CD4 ϩ T cells were isolated from singlecell suspensions, and cytosolic and nuclear extracts were prepared. 2.5 and 20 l of nuclear and cytosolic extracts, respectively, were added to ELISA plates with pre-bound B-specific oligonucleotide according to the instructions of the manufacturers. The plates were probed with polyclonal rabbit sera against p65 or p50; after development the absorbance was read. A, p65 and p50 levels in cytosolic and nuclear extracts of SEA-tolerized and -activated T cells. B, verification of binding specificity. Nuclear extracts prepared 1 h after the last SEA injection from SEA-activated (p65) and -tolerized (p50) CD4 ϩ T cells were incubated either alone or in combination with a mutated B oligonucleotide or a wild-type B oligonucleotide. HeLa extract and no extract were used as positive control and blank, respectively. S.D. were less than 10% of the mean. One of three similar experiments is shown.
cooperates with p50 to inhibit transcriptional activity of the IL-2 gene.

DISCUSSION
In this report, we analyzed the molecular alterations involved in the perturbed expression of NFB in tolerant CD4 ϩ T cells. We demonstrate that a change in composition of NFB complexes binding to the DNA in tolerant T cells resulted in complete suppression of NFB-driven transcriptional activity and IL-2 protein production. NFB binding to the B site of the IL-2 promoter was dominated by p50-p50 homodimers in tolerant CD4 ϩ T cells, whereas activated T cells primarily contained p65-p50 heterodimers. Evidently, the ratio between these different NFB complexes is important for the biological outcome of the immune response. Consistent with this, increased IL-2B-dependent transcription has been associated with increased p65-p50 heterodimer and decreased p50-p50 homodimer expression upon stimulation of CD4 ϩ T cell clones (36). Transfection of T lymphoma cells with p50 also revealed that increasing the amount of p50 will decrease the response of the IL-2 promoter in a dose-dependent manner (36). Moreover, reduced expression of tumor necrosis factor-␣ by tolerant macrophages has been associated with binding of p50-p50 homodimers to the B3 element of the mouse TNF-␣ promoter (20). Interestingly, macrophages from p50Ϫ/Ϫ mice were resistant to tolerance induced by LPS (20), and ectopic overexpression of p50 in the presence of constant amounts of p65 inhibited LPS-induced transcription of the TNF-␣ promoter, thus mimicking LPS-induced tolerance (20). LPS-induced tolerance in human monocytes is also associated with a predominance of p50-p50 homodimers binding to the TNF-␣ promoter, resulting in reduced TNF-␣ production (20,37). However, the p65 protein was translocated into the nucleus of tolerant cells as efficiently as in naive cells, suggesting that the degradation of IBs was intact (37). Hence, a nuclear overexpression of p50 can block p65-driven transactivation even in the presence of mobilized p65 (37). Taken together, these findings emphasize the importance of the p50 protein in the negative regulation of transcription.
Interestingly, we detected p65-and p50-containing NFB complexes without association of the c-rel component, although c-rel may be important in NFB-specific regulation of the IL-2 gene (38,39). Whereas TCR and CD28 receptor-mediated T cell activation will result in an early translocation of p65 to the nucleus, c-rel is translocated during a later phase of the immune response (40). The induction of such a biphasic response would change the composition of NFB complexes in the nucleus during T cell activation (40 -43). Because our studies were performed ex vivo, the lack of c-rel activation may be because of different kinetics of IL-2 production upon in vivo activation compared with naive T cells stimulated in vitro.
Upon SEA treatment, p65 was rapidly translocated from the cytosol into the nucleus of activated T cells, whereas the mobilization of p65 was strongly inhibited in tolerant T cells, most likely because of the poor activation of IKK and subsequent degradation of the IBs. The transient degradation of IB␣ early in the response of tolerant T cells correlated with the simultaneous appearance of nuclear p65. Whereas the activation-induced degradation of IB␤ was almost absent in tolerant FIG. 5. Poor phosphorylation and degradation of IBs in tolerant CD4 ؉ T cells. TCR V␤3 transgenic mice were injected one or three times i.v. with 10 g of SEA at 4-day intervals. Spleens and lymph nodes were removed at the indicated periods of time after the last SEA injection; cytosolic extracts were prepared from ex vivo purified CD4 ϩ T cells. A, cytosolic extracts were immunoblotted with antibodies against IB␣, IB␤, or ␤-actin. B, cytosolic extracts were first immunoblotted with an IB␣ antibody. After stripping, the membrane was reblotted with a p-IB␣ antibody. One of five similar experiments is shown.
FIG. 6. IKK activation is defective in in vivo tolerized CD4 ؉ T cells. TCR V␤3 transgenic mice were injected one or three times i.v. with 10 g of SEA at 4-day intervals. Spleens and lymph nodes were removed 24 h after the last SEA injection; purified CD4 ϩ T cells were prepared for in vitro cultures. T cells (10 ϫ 10 6 /time point) were put on ice (0 min) or stimulated with SEA (1 g/ml) and DCs (1 ϫ 10 6 /time point) for the indicated periods of time. After stimulation, cellular extracts were prepared and assayed for IKK activity using an immunocomplex protein kinase assay analyzing phosphorylation of GST-IB␣ and IKK autophosphorylation. Radioactive proteins were separated by SDS-PAGE, dried onto a filter, and visualized by autoradiography. Abundance of IKK␣ and -␤ proteins was analyzed by Western blotting. One of two similar experiments is shown. T cells, it followed slower kinetics compared with IB␣ in activated T cells, as previously described (16). Hence, the greatest difference in NFB activity between tolerant and activated T cells may be at later time points, when the IB␣ protein is back to control levels and when the sustained NFB response in activated T cells is due to degradation of IB␤.
The general overexpression of p50 in tolerant compared with activated T cells could be because of de novo synthesis of p50. However, the mRNA levels of the precursor protein p105 did not markedly differ between tolerant and activated T cells, which rather suggests post-translational cleavage of p105. Because the amount of p50 was considerably higher in 2ϫ SEA-treated resting T cells compared with untreated T cells, one can assume that de novo synthesis of p50 in tolerant T cells had already taken place even before the third SEA injection and thus preceding the true tolerant state. Several reports demonstrate that p50 is retained in the cytosol primarily through the formation of p105-p50 heterodimers, but the mechanisms involved in the processing of p105 to p50 are not fully understood (44 -46). However, it has been suggested that cotranslational as well as post-translational processes might be responsible for the generation of p50 from p105 (47).
The IB member, Bcl-3, is induced in T cells by mitogenic stimuli and in response to cytokines like IL-9 (24, 30). How-

FIG. 7. p105 is degraded only in activated T cells, whereas Bcl-3 is specifically induced in tolerant T cells.
TCR-V␤3 transgenic mice were injected i.v. one or three times with 10 g of SEA at 4-day intervals. Spleens and lymph nodes were removed at the indicated periods of time after the last SEA injection, and CD4 ϩ T cells were purified. A, total mRNA was isolated, and cDNA was prepared. Real-time PCR was performed on cDNA samples. Along with p105, Bcl-3, and IL-2, glyceraldehyde-3-phosphate dehydrogenase was also measured and used for compensation of unequal cDNA content. B, cellular extracts were prepared and immunoblotted with an antibody against p105 and p50. C, nuclear extracts were prepared and assayed for Bcl-3 association with B-specific oligos using ELISA. D, verification of binding specificity. Nuclear extracts prepared 1 h after the last SEA injection from SEA-tolerized CD4 ϩ T cells were incubated either alone or in combination with a mutated B oligonucleotide or a wild-type B oligonucleotide. HeLa extract and no extract were used as positive control and blank, respectively. S.D. were less than 10% of the mean. One representative experiment of two is shown. ever, even if it is clear that Bcl-3 preferentially binds to p50 or p52 homodimers, it is uncertain whether the Bcl-3-containing complexes promote or repress NFB-induced transcription. Several studies report an increased DNA-binding capacity of p50-p50 homodimers in the presence of Bcl-3 (28,30,44,48). Bcl-3 might then enhance transcription by formation of ternary complexes with p50-p50 or p52-p52 homodimers facilitating the recruitment of transcriptional co-activators and histone acetylases (49). Yet, it has also been demonstrated that the binding of Bcl-3-p50-p50 complexes to the DNA repress rather than transactivate NFB-dependent transcription (29,30,50,51). Whether the Bcl-3-p50-p50 complexes mediate positive or negative regulation may depend on the type of promoter or other cellular features (44). For instance, the location of Bcl-3 in the cytosol of hepatocytes was responsible for an autoregulatory feedback loop controlling the residence and action of nuclear p50-p50 homodimers (50). Because Bcl-3 is generally considered to be a nuclear IB protein, this kind of downregulation of NFB responses is likely to be cell type-specific.
To date, the involvement of Bcl-3 in T cell tolerance has not been described. Along with the high abundance of p50-p50 homodimers, we demonstrate here that tolerant CD4 ϩ T cells also expressed Bcl-3 mRNA and protein as well as its specific binding to the NFB site. It is possible that Bcl-3 is specifically induced in tolerant T cells to increase the DNA binding capacity of the p50-p50 homodimers in order to inhibit the transcriptional activity of the IL-2 gene, thus preventing the subsequent mitogenic response. However, even if Bcl-3-p50-p50 may be implicated in the inhibition of IL-2 transcription, it does not exclude the simultaneous induction of other genes in tolerant CD4 ϩ T cells. Recently it was shown that p50-p50 homodimers are responsible for the transcriptional induction of Bcl-2 in murine B cell lymphoma cells (52). A novel function of Bcl-2 is to inhibit apoptosis by the formation of a Bcl-2-p50 complex, which competes with active NFB in the nucleus and thus blocks apoptotic gene expression (53). In addition, p50-p50 homodimers are important in the transcriptional activation of the H-2K molecule (54).
Interestingly, Bcl-3 may be involved in the post-translational production of p50 and further favor the formation of p50-p50 homodimers. Signal-induced degradation of p105 could result from IKK-mediated phosphorylation and targeting by the proteasome (45,55,56); then released p50 might be sequestered and transported to the nucleus by Bcl-3. Furthermore, Bcl-3 may recruit p50 molecules from the cytoplasmic pool of p105-p50 heterodimers by a non-proteolytic mechanism (44). In any case, Bcl-3 would enhance the nuclear translocation and DNA binding of the p50-p50 homodimers. We found that, in contrast to activated T cells, tolerant T cells failed to activate the IKK complex. Additionally, the expression of p105 decreased upon stimulation in activated T cells, possibly as a result of p105 degradation. It is tempting to speculate that there might be an IKK-mediated degradation of p105 in activated T cells, whereas there may be a non-proteolytic formation and nuclear translocation of p50-p50 homodimers in tolerant T cells.
In summary, we demonstrate that the qualitative difference in the composition of NFB complexes binding to the IL-2 promoter in tolerant CD4 ϩ T cells correlated with reduced transcriptional activity and IL-2 production. The defective nuclear translocation of the p65 subunit is likely the result of reduced activation of the IKK complex and, thus, inhibited phosphorylation and degradation of the IB proteins retaining p65-p50 heterodimers in the cytosol. The induction of Bcl-3 in tolerant T cells may be relevant in the nuclear shuttling of p50-p50 homodimers, which could be the result from a nonproteolytic recruitment of p50 from p105 heterodimers in the cytosol. The simultaneous expression of Bcl-3 and p50 in the nucleus of tolerant T cells might increase the DNA binding capacity of p50-p50 homodimers and thus inhibit the binding and transcriptional activity of active NFkB heterodimers. This is supported by our observation that the B consensus site recruited Bcl-3-and p50-containing complexes. The differential regulation of p50 production and NFB dynamics in tolerant versus activated T cells may represent physiologically important ways of transcriptional activation and repression. We conclude that the impaired activation of NFB in tolerant T cells is a result of both passive and active repressor signals contributing to the regulation of NFB at multiple levels.