CD4+ T cells with a block in the NF-κB signaling pathway exhibit decreases in Th1 responses and diminished nuclear levels of multiple transactivating NF-κB/Rel/IκB proteins. To determine the lineage-intrinsic contributions of these transactivators to Th differentiation, T cells from mice deficient in specific subunits were cultured in exogenous cytokines promoting either Th1 or Th2 differentiation. RelB-deficient cells exhibited dramatic defects in Th1 differentiation and IFN-γ production, whereas no consistent defect in either Th1 or Th2 responses was observed with c-Rel-deficient cells. In sharp contrast, Bcl-3-null T cells displayed no defect in IFN-γ production, but their Th2 differentiation and IL-4, IL-5, and IL-13 production were significantly impaired. The absence of RelB led to a dramatic decrease in the expression of T-box expressed in T cells and Stat4. In contrast, Bcl-3-deficient cells exhibited decreased GATA-3, consistent with evidence that Bcl-3 can transactivate a gata3 promoter. These data indicate that Bcl-3 and RelB exert distinct and opposing effects on the expression of subset-determining transcription factors, suggesting that the characteristics of Th cell responses may be regulated by titrating the stoichiometry of transactivating NF-κB/Rel/IκB complexes in the nuclei of developing helper effector cells.
The diversity in pathogen characteristics dictates that acquired immune responses need to be able to adapt their characteristics according to the nature of the challenge. To meet this need for flexibility, CD4+ T lymphocytes, upon TCR ligation, hold the potential to develop into at least two functionally distinct effector populations, termed Th1 and Th2 cells. Specific functions of Th1 and Th2 effector cells are mediated through the expression of distinct arrays of cytokines. Characterized by the production of IFN-γ and TNF-β, Th1 cells are essential for host defense against intracellular pathogens, mediate delayed-type hypersensitivity responses, and have been associated with certain autoimmune diseases. Th2 cells, through the expression of IL-4, IL-5, IL-9, and IL-13, orchestrate host defense against helminthic infections, modulate Th1-mediated inflammation, and are central to the pathophysiology of atopic and allergic diseases. Due to these central roles in determining the characteristics of immunity, uncovering the mechanisms involved in this highly regulated process remains a focus of intensive study. Factors that influence whether a newly activated T cell differentiates into either a Th1 or Th2 effector include the type of Ag and its concentration, the type of cell presenting the Ag, and the cytokine milieu present at the time of Ag stimulation (1, 2). Each of these factors then modulates the levels of transcription factors induced in the activated CD4+ T cell.
Because it represents a fundamental paradigm in cellular differentiation and commitment, particular effort has been directed at identifying transcription factors that regulate the Th1/Th2 differentiation process and characterizing how they participate in the control of gene expression. To date, several transcription factors have been identified as having T cell-intrinsic roles critical to the Th1/Th2 differentiation process. c-Maf, NFATc, Stat6, and GATA-3 are all factors identified to be involved in the development of Th2 responses, whereas Stat4, Stat1, T-box expressed in T cells (T-bet), 4 and IFN regulatory factor-1 have roles in the development of Th1 responses (3, 4). Because all evidence indicates that differentiation proceeds in each mature CD4+ T cell starting from an uncommitted state, it is likely that the induction of a subset-specific factor, T-bet or GATA-3, is regulated by factors that are present in uncommitted T cells and are not subset specific.
NF-κB/Rel transcription factors are a family of evolutionarily conserved and structurally related proteins that are ubiquitously expressed in mammalian cells and play a central role in innate immune responses (5). The five known mammalian NF-κB/Rel proteins: p65 (RelA), RelB, c-Rel, p50 (NF-κB1), and p52 (NF-κB2) can be found in multiple combinations of homo- and heterodimers. Dimers containing RelA, RelB, or c-Rel are able to mediate transactivation directly (5). However, because neither the p50 nor p52 subunits possess transactivation domains, homodimers of these proteins act as repressors of gene transcription (5). NF-κB dimers are held in the cytoplasm by a class of related proteins called IκB. After the appropriate stimulation, the typical forms of IκB proteins are phosphorylated by the IκB kinase complex, polyubiquinated, and degraded by the 26S proteasome. Degradation of IκB then allows NF-κB dimers to translocate to the nucleus. Bcl-3 is a member of the IκB family that exhibits unique properties compared with the other family members, which include IκBα, IκBβ, IκBε, IκBg, and IκBz. Unlike other IκB proteins, which are predominantly located in the cytoplasm, Bcl-3 is largely nuclear (6) and can foster the transactivation of NF-κB-responsive genes in part by an indirect mechanism by which Bcl-3 displaces p50 or p52 homodimers from NF-κB consensus sequences (7, 8). This action relieves the transrepressive activity mediated by p50 or p52 homodimers and allows NF-κB dimers with the capacity for transactivation to bind and mediate gene expression. Furthermore, Bcl-3 contains a transactivation domain and can directly transactivate NF-κB-responsive genes when complexed with either p50 or p52 homodimers (9, 10, 11).
We recently determined that T cells expressing a degradation-resistant form of IκBα, called IκBα(ΔN), exhibited defective Th1 responses both in vivo and in vitro. The defect resulted from deficiencies in survival, clonal expansion, differentiation into Th1 effectors, as well as a block of postdifferentiation IFN-γ production (12, 13). In parallel, studies of NF-κB1-deficient animals showed that GATA-3 expression and Th2 cytokine production by NF-κB1 (p50)-deficient T cells were reduced, suggesting that NF-κB1 may be necessary for the development of Th2, but not Th1, responses (14). To dissect the role of NF-κB in the development of Th1 and Th2 responses, we investigated the contributions made by three transactivating members of the NF-κB/Rel/IκB signaling axis, c-Rel, RelB, and Bcl-3, to the development of effector T cell responses. We show that although RelB is necessary for optimal expression of T-bet and Th1 differentiation, Bcl-3 is required for normal GATA-3 levels and Th2 responses.
Materials and Methods
The mice used in this study were wild-type (WT) D011.10 αβ TCR transgenic (15), D011.10/IκBα(ΔN) double-transgenic (16) (both on a B10. D2 background), and C57BL/6 mice with targeted deletion of Bcl-3 (17), c-Rel (18), or RelB (19). All mice were maintained at the Vanderbilt University Medical Center mouse facility in specific pathogen-free conditions using microisolator cages and were used in accordance with applicable regulations after institutional approval. Experiments were generally performed using mice in the age range of 3–6 wk.
Cell culture conditions
Spleens from WT mice or mice deficient in Bcl-3, RelB, or c-Rel were harvested and dispersed into single-cell suspensions in complete IMDM (Invitrogen Life Technologies) supplemented as previously described (13). Splenic T cells were activated with plate-bound Abs against CD3ε and CD28 (BD Pharmingen) as described previously (16). Enriched populations of CD4+ T cells were generated by negative selection of MHC class II+ and CD8+ cells as described previously (13) to >95% purity. For the development of Th1-polarized populations, mIL-12 (10 ng/ml; Leinco Technologies), anti-IL-4 (1 μg/ml; BD Pharmingen), and rIL-2 (a gift from Biologic Response Modifiers Program) were added. For initiation of Th2-polarized cells, mouse IL-4 (10 ng/ml), anti-IL-12 (1 μg/ml), anti-IFN-γ (1 μg/ml), and human rIL-2 were used. For restimulation, cells were rinsed and equal numbers of viable cells were restimulated with immobilized Abs against CD3ε and CD28. After 48 h of restimulation, supernatants were harvested, and cytokine production was determined by ELISA. The production of IFN-γ, IL-4, and IL-5 was measured using BD Pharmingen ELISA Ab pairs. IL-13 was measured with reagents from R&D Systems. ELISAs were performed according to manufacturer’s instructions. Statistical analysis (unpaired two-tailed Student’s t test) was performed with INSTAT software (GraphPad). Over the course of conducting the experiments presented in this report, several different company-purchased lots of cytokine standards (IFN-γ and IL-4) were used. Because the absorbance units per nanogram of reference protein from these different lots varied from lot to lot, all cytokine ELISA data are presented in units per milliliter to allow more rigorous standardization and cross-comparisons.
For intracellular cytokine staining, cultured cells were rinsed and restimulated with PMA (50 μg/ml) and ionomycin (1 μg/ml; Sigma-Aldrich) for a total of 4 h or with Abs against CD3 (1 μg/ml) and CD28 (1 μg/ml) for a total of 6 h. Monensin (eBiosciences) was added for the last 3 h. Surface immunofluorescent staining was performed using allophycocyanin-conjugated anti-CD4 or biotinylated anti-CD4, followed by streptavidin-PerCP, after which cells were washed and fixed overnight in 4% paraformaldehyde in PBS at 4°C. The following day cells were permeabilized with 0.1% saponin and stained intracellularly with FITC-, allophycocyanin-, or PE-conjugated Abs against IFN-γ or IL-4. Intracellular levels of T-bet and GATA-3 were determined by incubating cells with purified mouse IgG1 (isotype control; BD Pharmingen), monoclonal anti-mouse T-bet (clone 4B10), or monoclonal anti-mouse GATA-3 (Santa Cruz Biotechnology), followed by PE-conjugated rat anti-mouse IgG1 (BD Pharmingen). Data were acquired using a FACSCalibur flow cytometer and CellQuest software (BD Biosciences), gating on forward and side scatter properties of viable lymphoid cells. Analysis was performed with FlowJo (TreeStar).
Cell extracts and immunoblot assay
Cytoplasmic and nuclear extracts were prepared by incubating cells first with buffer A (10 mM HEPES, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.4% Nonidet P-40, 1 mM DTT, and 0.5 mM PMSF) to generate cytoplasmic extracts, and then with buffer B (20 mM HEPES, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF, and 1% mammalian protease inhibitor mixture) to generate nuclear extracts. Whole cell extracts were generated as described previously (16). For detection of phosphorylated proteins, Na3VO4 and NaF were added to the lysis buffers. Western blot assays were performed with primary Abs against RelA, c-Rel, RelB, p50, Bcl-3, GATA-3, T-bet, Mel-18, Stat1, and Stat6 (all purchased from Santa Cruz Biotechnology). Anti-phospho-Stat6 (Y641) and anti-phospho-Stat1 (Y701) were purchased from Cell Signaling Technology, anti-phospho-Stat4 (Y693) and anti-Stat4 from Zymed Laboratories, and anti-cyclophilin B from Alexis Biochemicals.
For EMSAs D011.10 TCR transgenic splenocytes were stimulated with OVA323–339 overnight. Nuclear extracts were prepared, and EMSAs were performed as described previously (13). Supershift EMSA was performed by preincubating extracts with 2 μl of anti-p50 (Santa Cruz Biotechnology) for 20 min before addition of end-labeled probe. For competition EMSA, excess unlabeled oligonucleotides were preincubated with extracts before addition of radiolabeled probe. The sequences of the synthetic nucleotides used for EMSA and competition experiments were: κB-GATA-3, 5′-AGA GAG CGA ATT CCC TCC TGC CTC T-3′; κB-palidromic oligo, 5′-CAA CGG CAG GGG AAT TCC CCT CTC CTT-3′; Sp1, 5′-ATT CGA TCG GGG CGG GGC GAG C-3′; and AP-1 5′-CGC TTG ATG AGT CAG CCG GAA-3′. For oligonucleotide pull-down assays, whole cell protein extracts were prepared from ΦNX cells transfected with pCMV4-p50, MSCV-IRES-Thy1.1-Bcl-3, or pCMV4-p65 expression constructs as previously described (13) and were incubated with a biotin-labeled κB-GATA-3 double-stranded oligonucleotide (30 min at 4°C) and then with streptavidin-agarose beads for an additional 30 min (Novagen). Washed beads were boiled in SDS-PAGE buffer, and the DNA-associated NF-κB proteins were resolved on SDS-PAGE gels and detected by immunoblot assay.
DNA constructs and reporter assays
Jurkat T cells, maintained in complete RPMI 1640 supplemented with 5% FBS, were transfected via electroporation at a density of 10 × 106 cells/transfection in serum-free complete RPMI 1640 at 250 V and 960 mF. The constructs used in these experiments include MSCV-IRES-Thy1.1 (MIT), MSCV-IRES-Thy1.1-Bcl-3 (MIT-Bcl-3), pCMV4, pCMV4-p50, and a luciferase reporter plasmid κB-GATA-3 pBS-luc (a gift from Dr. I-C. Ho, Harvard University, Boston, MA) (20). Cell extracts were prepared in reporter lysis buffer (Promega), and luciferase activity was assayed according to the manufacturer’s instructions. For all luciferase assays, transfection efficiency was determined by staining a small aliquot of transfected cells from each condition with anti-Thy1.1-PE, and data were normalized by dividing raw light units by the percentage of Thy1.1-positive cells as determined by FACS.
Th1 responses of Bcl-3-deficient T cells are comparable to WT, but Th2 responses are impaired
Th1 responses are diminished in IκBα(ΔN) transgenic T cells compared with WT controls (12, 13). Moreover, T cells subjected to NF-κB blockade also exhibited a decrease in the expression of Bcl-3 (13). This observation raised the possibility that Bcl-3 contributes to the development of Th1 responses. Therefore, we compared the amounts of cytokine produced by WT and Bcl-3-deficient T cells after in vitro differentiation under conditions promoting Th1 and Th2 development followed by secondary stimulation. IFN-γ production by Bcl-3-deficient T cells was not significantly different from that by WT controls (Fig. 1,A). In contrast, expression of the hallmark Th2 cytokine IL-4 was significantly diminished. Intracellular staining for IL-4 indicated that these defects represented an impairment of Th2 differentiation in Bcl-3-deficient cells. The percentage of IL-4-positive cells in the Bcl-3-null Th2 cultures was only 28% that in WT controls, whereas no consistent change in the percentage of IFN-γ-positive cells was detected in the Th1 cultures (Fig. 1,B). The defect was not restricted to IL-4, in that production of other Th2 cytokines (IL-5 and IL-13) was also diminished (Fig. 1 C). These data, obtained when a sufficient level of differentiating cytokine (IL-12 or IL-4) was present and antagonistic cytokine action was blocked (anti-IL-4 or anti-IL-12, respectively), indicate that Bcl-3 plays a T cell-intrinsic role in type 2 Th differentiation and cytokine release.
Nuclear expression of RelB and c-Rel is decreased within IκBα(ΔN) T cells
Because experiments with Bcl-3-deficient cells revealed that Bcl-3 within T cells is not essential for the development of Th1 responses, we sought to determine contributions by other transactivating subunits. We had previously shown that an activation-induced increase in Bcl-3 levels depended on the canonical NF-κB pathway (13), so we began by investigating how T cell-specific expression of the IκBα(ΔN) transgene affected nuclear levels of RelA, c-Rel, and RelB in T cells. WT and IκBα(ΔN) cells were stimulated with Ag, and the cytoplasmic and nuclear levels of RelA, c-Rel, and RelB were assayed by immunoblot. Although the overall expression of each subunit was not substantially altered, the nuclear levels were decreased in IκBα(ΔN) cells 24 h (Fig. 2) and 72 h (data not shown) after activation. These data indicated that nuclear levels of RelA, c-Rel, and RelB are reduced in the transgenic T cells, suggesting that these transactivating Rel family members might contribute to the development of Th1 responses.
Decreased Th1 differentiation and IFN-γ production by RelB-deficient T cells
RelA deficiency leads to embryonic lethality, whereas c-Rel- and RelB-null mice are viable and exhibit normal T cell development (5, 18, 19). Therefore, we cultured T cells deficient in either c-Rel or RelB under Th1- and Th2-polarizing conditions and assayed the levels of IFN-γ and IL-4 after restimulation. With the inclusion of sufficient IL-2 (100 U/ml) to the cultures, T cells lacking c-Rel did not exhibit a consistent deficit in the production of either IFN-γ or IL-4 (Fig. 3,A). In contrast, a significant and consistent decrease in the production of IFN-γ, but not IL-4, was detected in cells deficient in RelB (Fig. 3,B). In line with previous findings using IκBα(ΔN) T cells (13), the growth of RelB-deficient T cells was impaired (data not shown). However, because the production of IFN-γ was assayed upon secondary stimulation of equal numbers of RelB−/− cells and controls, this finding represented a deficit in IFN-γ production, rather than a paucity of cells. Importantly, intracellular cytokine staining data showed that the defect in IFN-γ production was attributable to inefficient differentiation, because a much lower fraction of IFN-γ-positive cells was detected among the RelB-null CD4+ T cells cultured in Th1-differentiating conditions compared with WT cells (Fig. 3,C). To address the possibility that the defect in Th1 differentiation was due to a lack of some factor derived from any APCs in the system, CD4+ T cells deficient in RelB were primed under Th1 conditions in the presence of WT APC. T cells lacking the expression of RelB remained defective in the Th1 response despite provision of normal APCs, supporting the idea that the defect is intrinsic to the T cell (Fig. 3 D). Together, these data indicate that although a deficiency in c-Rel did not lead to a consistent defect, RelB in T cells influences Th1 development.
Diminished expression of T-bet and Stat4 in RelB-deficient Th1 cells
RelB-deficient Th1 cells, like IκBα(ΔN) cells, exhibited a severe deficit in IFN-γ production. Previously, we determined that there was a modest diminution in the levels of T-bet expressed in IκBα(ΔN) Th1 cells relative to WT controls (13). When WT and RelB-deficient Th1 cells were compared at a time point during the differentiation process, a striking defect in T-bet expression was detected in cells lacking RelB (Fig. 4,A). Moreover, the intracellular levels of T-bet detected in the RelB-deficient cells cultured in the presence of IL-12 were decreased compared with WT controls (Fig. 4,B). T-bet is up-regulated through the IFN-γ/Stat1 signaling axis (21) so that T-bet expression might be enhanced in an autocrine fashion as IFN-γ was released by differentiating cells. This raised the possibility that the decrease in T-bet induction was an indirect consequence and due to an initial paucity of IFN-γ in the RelB-deficient Th1 cultures. To investigate this possibility, two types of control analysis were performed. In one, we compared the levels of T-bet expressed by WT cells differentiated under Th1 culture conditions in the presence or the absence of neutralizing anti-IFN-γ (to simulate the effect of reduced IFN-γ during the priming). In a complementary analysis, T-bet was measured in RelB knockout (KO) cells differentiated under Th1 conditions in the presence or the absence of exogenous IFN-γ (Fig. 4 C). The defect in T-bet induction exhibited by RelB-deficient T cells was not reversed by provision of IFN-γ; conversely, blocking IFN-γ led to, at most, a modest reduction in T-bet.
Because T-bet is thought to be downstream from Stat1, it was possible that the exogenous IFN-γ in RelB KO cultures did not alter T-bet expression, because IFN-γ signaling to Stat1 was defective in these cells. However, comparison of RelB-deficient T cells to controls showed, at most, a modest reduction in Stat1 expression and IFN-γ-induced phosphorylation of Stat1 (Fig. 4,D) to an extent unlikely to account for the dramatic defect in T-bet expression. These findings demonstrate that the observed decrease in T-bet is an intrinsic one, due primarily to the absence of RelB rather than to either an initial decrease in the amount of IFN-γ, which was essential for driving a feed-forward amplification loop, or an inability to activate the Stat1 pathway. Th1 differentiation also involves a substantial increase in levels of total Stat4 protein (22), and we have previously shown that the levels of Stat4 are significantly decreased in IκBα(ΔN) Th1 cells as compared with WT (13). In a manner paralleling these earlier findings, the overall level of Stat4 as well as the amount of phosphotyrosyl-Stat4 were significantly decreased in developing Th1 cultures lacking RelB (Fig. 4 E). Together, these data suggest that RelB/NF-κB acts at a point upstream of T-bet and Stat4 to influence the development of Th1 effectors and also indicate that RelB and Stat1 may collaborate in mediating T-bet induction.
Bcl-3-null Th2 cells have a defect in GATA-3 expression that is independent of Stat6 or Mel-18
GATA-3 is preferentially expressed in mature Th2 cells and is thought to facilitate global chromatin remodeling and activation of the IL-4/5/13 locus, thereby regulating the expression of these Th2 cytokines (23, 24, 25). After initial Th2 differentiation, loss of GATA-3 appears not to affect IL-4 production (26). Because Bcl-3 KO Th2 cells exhibited significant decreases in the production of multiple Th2 cytokines, we investigated the expression of GATA-3 by these cells at a time point before completion of differentiation in the cultures. GATA-3 expression in Th2 conditions was dramatically decreased in nuclear extracts of Bcl-3-null cells compared with that in WT cells (Fig. 5,A). In contrast, the expression of T-bet was unaffected by the absence of Bcl-3 under Th1 conditions (Fig. 5,A). GATA-3 expression is specific to Th2 cells, and Bcl-3 KO cultures had a lower fraction of Th2 cells (Fig. 2). Therefore, it was possible that the decrease in GATA-3 was a consequence of a lower frequency of turning on GATA-3 to generate IL-4-positive cells in the KO Th2 cultures. To determine whether the decrease in this transcription factor was exclusively a consequence of fewer IL-4-positive cells being present in the Bcl-3 KO Th2 cultures, or whether the expression of GATA-3 within each Bcl-3-null cell was decreased, intracellular FACS for GATA-3 expression was performed. Bcl-3-deficient CD4+/IL-4+ cells exhibited a modest, but consistent, reduction in intracellular GATA-3 levels compared with WT CD4+/IL-4+ cells (Fig. 5,B). A similar trend was exhibited by the uncommitted CD4+/IL-4−/IFN-γ− population (Fig. 5 B). We conclude that Bcl-3 expression within differentiating Th2 cells is critical for achieving full induction of GATA-3 expression, and that the overall decrease in GATA-3 levels observed in Western blots is a consequence of both this intrinsic defect and decreases in the relative numbers of IL-4-positive cells in Bcl-3 KO Th2 cultures.
Under cytokine-driven Th2 conditions, the induction of GATA-3 is dependent on the expression and activation of Stat6 as well as the polycomb group gene Mel-18. T cells from mice rendered deficient in either Stat6 or Mel-18 exhibit decreased ability to produce IL-4, IL-5, and IL-13 as well as a decrease in the expression of GATA-3 (27, 28), and the promoter of Mel-18 has been reported to contain a NF-κB like binding site (29). Therefore, we tested whether the defect in GATA-3 involved defects in the expression/activation of Stat6 or expression of Mel-18. In the absence of Bcl-3, both expression and IL-4 induction of Stat6 were normal (Fig. 5,C). Furthermore, no defect in the expression of Mel-18 was detected in cells lacking Bcl-3 compared with WT (Fig. 5 D). We conclude that GATA-3 induction is impaired in Bcl-3 null cells despite apparently normal IL-4/Stat6 signaling and Mel-18 expression.
Bcl-3 and NF-κB1 p50 binding to and transactivation of the murine gata-3 promoter
GATA-3 expression is also defective in Th2 cells that are deficient in NF-κB1 p50 (14), a DNA-binding partner of Bcl-3. A κB-like consensus sequence can be identified at position −310 to −301 within one of the murine gata-3 promoters (Fig. 6,A), although whether it can bind NF-κB proteins is not clear. Together with our findings, these points suggested that NF-κB might participate directly in the induction of GATA-3 expression. Mobility shift assays showed that proteins in nuclear extracts from antigenically stimulated T cells bound to the gata-3 promoter κB site to generate complexes comigrating with NF-κB (Fig. 6,B). Similar complexes were formed using extracts from cells transfected to overexpress NF-κB (data not shown). Association of NF-κB with this gata-3 κB-oligo is specific, because the shift can be competed away with an oligo known to be specific for NF-κB (13) (Fig. 6,B, lane 4), but not with oligos specific for Sp1 or AP-1 (Fig. 6,B, lanes 5 and 6). Furthermore, the shifted band could be supershifted with anti-p50 Abs, indicating that p50 is present in these complexes (Fig. 6,C). Such a supershift was not reproducibly generated with anti-Bcl-3, perhaps due to limitations of the available Abs, so we used oligonucleotide affinity chromatography to determine whether Bcl-3 also associated with this gata-3 promoter κB site. Using extracts of cells expressing high levels of p50, Bcl-3, or both, immunoblots of proteins eluted after binding to a biotinylated κB oligo demonstrated that Bcl-3 is in these complexes and confirmed that p50 associates with this κB site (Fig. 6,D). Because Th2 cells deficient in Bcl-3 have decreased levels of GATA-3, we next sought to determine whether the association of Bcl-3 with this region of the murine gata-3 promoter has functional relevance. Using a reporter construct that contained this murine gata-3 κB consensus sequence (Fig. 7,A), we found that coexpression of Bcl-3 and p50 significantly transactivated the promoter (Fig. 7, B and C), strongly suggesting that Bcl-3, along with its DNA-binding partner, p50, participates in the transactivation of gata-3.
The immune system exhibits the capacity to adjust the balance of effector molecules as a consequence of differences in the conditions during a response. Thus, antigenic stimulation concurrently elicits the differentiation of several functionally distinct subsets of Th cells from the initial pool of Ag-reactive activated T cells (32), but the strength of each limb of the Th response is modulated differently when comparing one immune challenge with another. The ultimate differentiation of an activated CD4+ T cell into the Th1 or Th2 subset appears to be driven by subset-specific master regulator proteins, such as T-bet and GATA-3 (3, 33). Accordingly, a fundamental challenge in understanding how the characteristic balance of Th1 and Th2 cells can be achieved in response to differing circumstances is to determine which transcriptional pathways influence the induction of these subset-specific regulators. Previous evidence has suggested that Notch signaling and the NFAT family proteins may play Janus-like roles in the development of Th1/Th2 responses (34, 35, 36, 37, 38). The central finding of the present work is that specific transactivating members of the NF-κB/Rel/IκB family of transcription factors play opposing roles in regulating the induction of T-bet and GATA-3, leading to differential effects on Th1 and Th2 development. These data serve to resolve a potential paradox between findings for IκBα(ΔN) transgenic mice, in which NF-κB was found to be preferentially important to Th1 responses, and NF-κB1/p50 knockouts, in which defects in Th2 responses were observed (12, 13, 14). Furthermore, the findings presented in this report suggest that physiologically mediated shifts in the balance between Bcl-3 and RelB in the nuclei of activated CD4+ T cells might regulate the balance between Th1 and Th2 responses.
The Th1 response to an intracellular pathogen (39) or defined Ags (12, 13) was dramatically deficient in studies using a transgenic system in which the ability of T cells to signal through NF-κB/Rel/IκB pathways was specifically inhibited only in T cells. This outcome arose due to impairments in the efficiency of Th1 differentiation as well as dramatic defects in clonal expansion and postdifferentiation IFN-γ production (13). These effects were accompanied by decreased induction of the transactivating IκB-like protein, Bcl-3, in transgenic T cells (13). In the present study, however, Bcl-3-null T cells exhibited a selective defect in their production of type 2 cytokines and normal IFN-γ production. We infer from this finding that the impaired induction of Bcl-3 does not account for the defect in Th1 differentiation observed when the canonical NF-κB pathway is blocked in T cells. Bcl-3-deficient mice were more susceptible to Toxoplasma gondii infection in association with a diminished capacity of T cells to produce IFN-γ (17). Taken together with our present findings on the intrinsic capacity of Bcl-3-deficient T cells to differentiate into Th1 cells and produce IFN-γ upon secondary TCR stimulation, the data suggest that the previously observed defects are due to T cell-extrinsic factors, such as altered lymphoid architecture or APC function (17), or may reflect outcomes dependent on the nature of the Ag/infection.
Bcl-3 deficiency results in phenotypes that in many ways parallel those observed with NF-κB1 deficiency (40, 41). Mice with targeted deletion of NF-κB1 have been shown to have defects in both type 1 and type 2 responses (14, 42, 43), and NF-κB1 KO T cells exhibited decreased GATA-3 levels and production of Th2 cytokines when differentiated in vitro under conditions similar to those used in the present study. The diminished capacity of Bcl-3-deficient cells to develop into Th2 cells involved a defect in GATA-3 expression that was independent of Stat6 or Mel-18. We infer that, as for p50-null cells, Bcl-3 influences Th2 development at least in part through its cooperation with other factors (such as Stat6) in the regulation of GATA-3. Bcl-3 transactivates NF-κB-responsive genes when complexed with p50 dimers (10), and NF-κB-like sites have been identified in the putative gata-3 promoter (14, 30). Our data provide the first evidence that a NF-κB/Rel complex, containing p50 and Bcl-3, can bind specifically to a κB-like site within one of the GATA-3 promoters and that combined p50/Bcl-3 can transactivate this promoter. Taken together, this evidence suggests that Bcl-3, in association with p50, regulates GATA-3 expression at least in part through direct transactivation. However, not all aspects of the mechanism by which Bcl-3 influences Th2 development are addressed by the data presented in this study. Because Bcl-3 has been shown to transactivate NF-κB-responsive genes when complexed with NF-κB1 dimers (9, 10), transactivation may be through a Bcl-3:p50:p50 complex. Alternatively, however, Bcl-3 may be promoting the removal of p50 homodimers from the κB-site, thus allowing a p50-containing heterodimer to interact with the κB site and transactivate gata-3 (7, 8). In addition, there are NF-κB-like sites in Th2 cytokine promoter DNA, and Bcl-3 is able to recruit complexes of potent coactivators. Thus, we cannot exclude the possibility that an additional role for Bcl-3 is through direct actions at the Th2 cytokine genes or that additional contributions are made by other factors (11, 44).
Although Bcl-3-deficient cells developed attenuated Th2 responses, RelB-deficient cells had a severe defect in Th1 differentiation and production of IFN-γ. Consistent with these findings, RelB-deficient mice were more susceptible to infection with T. gondii. Cells lacking RelB produced less IFN-γ immediately after in vitro stimulation (45), but this initial response is highly dependent on NKT cells, which have been found to be lacking in RelB−/− mice (46, 47). Although mice with targeted deletion of RelB present with impaired APC function and reduced numbers of myeloid-derived dendritic cells (19, 48), the population of lymphoid (CD8α+) dendritic cells, which promote Th1 differentiation, is maintained (49). This point and our data showing that culturing RelB-deficient T cells in the presence of WT APCs did not rescue the defect in Th1 responses indicate that defects in Th1 responses are not simply a reflection of a lack of suitable APCs. Moreover, the RelB-deficient T cells in the present work were stimulated directly via anti-TCR Abs and cultured in the presence of exogenous IL-12, so that defects related to defective Ag presentation or cytokine production would be bypassed. Roles for the IL-27 receptor, WSX-1, have recently been uncovered in early Stat1 induction and the Th1 response (50). However, WSX-1-deficient T cells, assayed under the conditions used in this study, develop normal Th1 responses (51), so that the findings with RelB-deficient T cells are not attributable to a defect in the IL-27/WSX-1 axis. Although it is known that c-Rel regulates IL-2 production by T cells and IL-12 production by APCs, investigation of the T lineage-intrinsic role of c-Rel indicated that these cells developed normal type 2 responses and exhibited no consistent defect in type 1 responses (Fig. 3). Other reports on the role of c-Rel in the Th1 response have yielded varying results both in vitro and in vivo (52, 53, 54, 55); even in the case where c-Rel influenced the amount of IFN-γ produced, the induction of T-bet was normal. Thus, c-Rel may in some conditions influence postdifferentiation IFN-γ production, a mechanism identified in dissection of the role of the canonical NF-κB pathway in Th1 differentiation vs postdifferentiation IFN-γ production after secondary stimulation (13). Taken together, these findings indicate that there is a T cell-intrinsic requirement for RelB during Th1 differentiation, whereas the role of c-Rel in the development of either Th2 or Th1 differentiation appears to be less robust.
The decreases in IFN-γ exhibited by RelB-deficient T cells coincided with decreases in the expression of T-bet and Stat4 (Fig. 5), findings that parallel our previous observations with IκBα(ΔN) T cells (13). Although T-bet is up-regulated through the IFN-γ/Stat1 axis (21), the defect in T-bet expression was not reversed by exogenous IFN-γ, and activation of Stat1 in RelB-deficient T cells is not significantly different from that in WT T cells, suggesting that RelB is in some way upstream from T-bet. Stat4 has also been reported to promote the induction of T-bet, so the dramatically decreased level of Stat4 protein evident in RelB-deficient cells under Th1 conditions suggests that, in part, the role of RelB in T-bet induction may be indirect. Moreover, Stat4 appears able to act in a T-bet-independent manner to induce the production of IFN-γ when developing Th2 cultures are switched to Th1 conditions (22, 56, 57). Although the mechanisms by which Th1 conditions lead to increased levels of Stat4 protein (22) have not been determined, the present data indicate that this process requires RelB. Thus, although additional mechanisms are presumably involved (e.g., actions independent of Stat4 or T-bet levels or direct involvement at the IFN-γ locus), the dramatic decrease in Stat4 in RelB−/− cells indicates that the significant defect in IFN-γ expression and T-bet involves regulation of Stat4 levels.
Specific NF-κB/Rel complexes can be differentially regulated in response to distinct immunologic stimuli and have been shown to exhibit distinct capabilities with regard to binding to and activating transcription from NF-κB consensus sequences (58, 59). Using T cells deficient in specific forms of NF-κB, we have shown that certain NF-κB/Rel/IκB proteins have divergent activities in Th cell differentiation. This conclusion is consistent with and may help explain recent data demonstrating that the consequence of increased NF-κB activity in T cells can be to enhance Th1 responses or, alternatively, Th2 responses. For example, activation of NF-κB is amplified in cells deficient in Foxj1, the consequence of which is an exaggerated production of Th1, but not Th2, cytokines (60). In contrast, Schnurri-2-deficient T cells have augmented activity of NF-κB, which was associated with increases in GATA-3 and development of Th2 cells (61). These differences in the effects of increased NF-κB activity may reflect differences in the precise complexes being activated. The findings presented in this study indicate that titration of RelB, which is heavily dependent on the alternative pathway for NF-κB induction, vs Bcl-3 might play a central role in regulating relative induction of T-bet and GATA-3 within each activated Th precursor cell.
We thank L. Glimcher for gifts of anti-T-bet Abs, T. Mitchell for MiT-Bcl-3; I. Ho for the GATA-3 pBS-luc construct; the Vanderbilt (J. Higginbotham) and Departmental Flow Cytometry Cores, L. Stephenson, S. Joyce, and T. Aune, for helpful discussions; and T. Dermody for critically reading the manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National Institutes of Health (NIH) Grant AI49460 with NIH support grants to Vanderbilt (Vanderbilt-Ingram Cancer Center Support Grant CA68485; Diabetes Center Grant P60 DK20593) and Department of Microbiology and Immunology, Vanderbilt University Medical School (P01 HL68744) core facilities. M.R.B. received additional support from NIH Grant HL68744 and the Sandler Program for Asthma Research; R.A.C. was supported by U.S. Department of Education Graduate Assistance in Areas of National Need P200A010123, an NIH award to the Meharry-Vanderbilt Alliance (R25 GM62459), and AI49460-02S1-04S1. Submitted in partial fulfillment of the requirements for a Ph.D. (to R.A.C., Meharry Medical College).
Abbreviations used in this paper: T-bet, T-box expressed in T cells; WT, wild type; KO, knockout.