We have investigated activation of nuclear factor-κB (NF-κB) in the process of primary B cell differentiation in vitro. In this system, NF-κB is strongly induced when B cells develop from the pre-B cell to the immature B cell stage. Unlike the typical NF-κB activation in response to exogenous stimuli, induction proceeds with a slow time course. NF-κB induction is only observed in B cells that undergo differentiation, not in Rag2-deficient cells. Nuclear DNA binding complexes predominantly comprise p50/RelA heterodimers and, to a lesser extent, c-Rel-containing dimers. The increase in NF-κB binding activity is accompanied by a slow and steady decrease in IκBβ protein levels. Interestingly, absolute RelA protein levels remain unaffected, whereas RelB and c-Rel synthesis is induced. The reason for preferential nuclear translocation of RelA complexes appears to be selective inhibition by the IκBβ protein. IκBβ can efficiently inhibit p50/RelA complexes, but has a much reduced ability to interfere with p50/c-Rel DNA binding both in vitro and in vivo. Interestingly, p50/RelB complexes are not at all targeted by IκBβ, and coimmunoprecipitation experiments show no evidence for an association of IκBβ and RelB in vivo. Consistent with these observations, IκBβ cotransfection can inhibit p50/RelA-mediated trans-activation, but barely affects p50/RelB mediated trans-activation.
NF-κB3 was originally discovered as a DNA binding activity that was constitutively present in mature B cells and plasma cells, but was present in a latent inducible form in pre-B cells and many other cell types (1, 2). Molecular cloning of genes encoding NF-κB proteins revealed the presence of a family of related proteins (the NF-κB/Rel family of transcription factors) that can bind to the κB motif as homo- and/or heterodimers (for recent reviews, see Refs. 3 and 4). To date, five family members have been identified and characterized in some detail: p50 (p105/NFKB1), p52 (p100/NFKB2), RelA, RelB, and c-Rel (5). The activity of these proteins is largely determined by specific inhibitor proteins, which are responsible for cytoplasmic retention and inhibition of DNA binding. Biochemical analyses and molecular cloning revealed two types of IκBs, designated IκBα and IκBβ (6, 7, 8). IκBε, a third related inhibitor protein was cloned very recently (9). In addition, the C-terminal domains of the NF-κB1 and NF-κB2 precursor proteins, which are proteolytically removed to generate the mature p50 and p52 proteins, respectively, have structural homologies with the IκB proteins and also function as inhibitors (10, 11). In some B cell lines, the C-terminal domain of NF-κB1 is indeed expressed as a separate IκB entity and was termed IκBγ (12). A further protein, Bcl3, can function both as specific inhibitor of p50 and p52 homodimer DNA binding as well as a transcriptional coactivator with these homodimers, presumably due to the formation of metastable ternary complexes present in the nuclei (13, 14, 15).
Mature B cell-specific and ubiquitous inducible NF-κB functions could recently be assigned to distinct members of this NF-κB/Rel family. Whereas RelA-containing complexes are typically seen upon induction of many different cell types by a variety of stimuli, the lymphoid-specific constitutive complexes predominantly contain RelB heterodimers (16, 17). Interestingly, c-Rel was shown to be constitutively present in several cell lines representing mature B cells or plasma cells, at a low level in primary spleen extracts, and also as an inducible species in B and T lymphoid cells (17, 18, 19, 20, 21, 22).
The molecular switch that accompanies pre-B to B cell maturation, which is associated with the occurrence of constitutively NF-κB binding proteins, has yet not been characterized. Several hypotheses have been put forward to explain constitutive vs inducible DNA binding activities. In addition to the preferential expression of RelB (and also c-Rel) in cells of the lymphoid lineage, reduced sensitivity of RelB to IκBα inhibition, which is most likely due to a lymphoid-specific modification of the RelB protein, has been suggested (23, 24). A further mechanism that could contribute to the constitutive presence of nuclear NF-κB binding proteins in mature B and plasma cells could be the increased turnover of IκBα protein in mature B cells compared with pre-B cells (21, 25). An alternative explanation for the low levels of nuclear NF-κB proteins in pre-B cells was suggested recently. Essentially all murine pre-B cell lines that had been characterized for the presence of NF-κB had been established by transformation with the Abelson retrovirus. Using a ts mutant of the Abelson virus, it was suggested that the abl-oncogene might directly or indirectly interfere with NF-κB activation (26). In support of this conclusion, they detected significant amounts of constitutive κB binding activity in primary lymphocyte cultures (Whitlock-Witte cultures). However, a disadvantage of this culture system is the fact that in addition to supporting the growth of pre-B cells, these cultures simultaneously allow differentiation of B cells and, therefore, represent a mixture of B cells representing different maturation stages (27, 28).
We have employed a more defined B cell culture and differentiation system that was developed recently (29). In the presence of IL-7 and stromal cells, pre-B cells can be continuously propagated. These cells typically have undergone DH-JH rearrangements on both alleles (30). After removal of IL-7 these cells continue heavy chain and light chain rearrangements and differentiate to surface Ig-positive, immature B cells. We used this system and showed that, like their transformed counterparts, primary pre-B cells have very low levels of nuclear NF-κB, demonstrating that these earlier findings were not only due to abl oncogene transformation. In addition, we used the IL-7 withdrawal/differentiation scheme to show that upon differentiation nuclear NF-κB levels increase with an unusually slow time course. Analysis of the steady state levels of IκB proteins during differentiation suggests that slow degradation of IκBβ contributes to the observed NF-κB increase.
Materials and Methods
Cell culture and transfections
The different pre-B cell lines were established and propagated in Iscove’s modified Dulbecco’s medium-based serum-free medium containing 100 to 200 U/ml IL-7 on gamma-irradiated ST-2 stromal cells (29). For initiation of the differentiation program, pre-B cells were washed three times in medium without IL-7 and then cultured as described above on stromal cells in medium lacking IL-7 for the time indicated. NIH-3T3 and COS cells were grown in DMEM supplemented with 10% FCS.
COS cells were transfected by electroporation (31) with 0.5 μg of p50 expression vector and 3 μg of RelA, RelB, or c-Rel expression vectors. NIH-3T3 cells were transfected by calcium phosphate coprecipitation using a total of 20 μg of DNA. Transfections contained 4 μg of luciferase reporter plasmid, 6 μg of NF-κB expression vectors (pRc/CMV based) as described in the figure legends, and 1, 3, or 9 μg of pRc/CMV-IκB expression vector (always brought to 9 μg with the empty expression vector). One microgram of a CMV-lacZ reporter was included in all transfections to normalize for differences in transfection efficiencies. Transfections were harvested after 24 h, and luciferase and β-galactosidase activities were determined (32).
The murine IκBβ cDNA was cloned by reverse transcription-PCR from a murine B cell line (S194) using primers deduced from the published IκBβ sequence (7). The cDNA was inserted into pRc/CMV (Invitrogen, San Diego, CA) and used for transfection experiments as well as in vitro transcription/translation.
For selection of B220-positive splenic B cells, hemolytic Gey’s solution was first used to remove RBC from a splenic cell suspension. After Gey lysis, splenic lymphocytes were centrifuged through a FCS cushion and washed with PBS, 5 mM EDTA, and 0.5% BSA. For sorting of B220-positive B cells a magnetic separation kit was used according to the manufacturer’s instructions (Miltenyi Biotec). Surface IgM-positive cells were selected from differentiated pre-B cell cultures after 3 days of IL-7 withdrawal. The cells were first passed over a Ficoll gradient to enrich for living cells and then were positively selected using rat anti-mouse IgM Abs coupled to magnetic beads (Miltenyi Biotec, Bergisch-Gladbach, Germany). Selected cell populations were analyzed by FACS using FITC-labeled rat anti-mouse IgM and phycoerythrin-labeled rat anti-B220 (PharMingen, San Diego, CA).
Cell extracts, conditions for EMSA, Western immunoblots, and immunoprecipitations
For preparation of nuclear and cytoplasmic extracts (33), pre-B cells were lysed in sucrose buffer I (100 μl/107 cells, 0.32 M sucrose, 3 mM CaCl2, 2 mM magnesium acetate, 0.1 mM EDTA, 10 mM Tris-HCl (pH 8.0), 1 mM DTT, 0.5 mM PMSF, and 0.5% Nonidet P-40). Nuclei were pelleted by centrifugation (500 × g, 5 min, 4°C), and 0.22 vol of 5× cytoplasmic extraction buffer (0.15 M HEPES (pH 7.9), 0.7 M KCl, and 0.015 M MgCl2) was added to the supernatant (=cytoplasm), microcentrifuged at 12,000 × g for 15 min at 4°C, and transfered to a fresh tube. Pelleted nuclei were washed in sucrose buffer I lacking Nonidet P-40, microcentrifuged (500 × g, 5 min, 4°C), and resuspended in low salt buffer (20 μl/107 cells, 20 mM HEPES (pH 7.9), 25% glycerol, 1.5 mM MgCl2, 0.02 M KCl, 0.2 mM EDTA, 0.5 mM DTT, and 0.5 mM PMSF). To extract nuclei, sequentially high salt buffer (20 mM HEPES (pH 7.9), 25% glycerol, 1.5 mM MgCl2, 0.8 M KCl, 0.2 mM EDTA, 1% Nonidet P-40, 0.5 mM DTT, and 0.5 mM PMSF) was added, incubated for 20 min on ice, diluted 1/2.5 with diluent (25 mM HEPES (pH 7.6), 25% glycerol, 0.1 mM EDTA, and 0.5 mM PMSF), and microcentrifuged (12,000 × g, 15 min, 4°C). Supernatant (=nuclear extract) was transfered to fresh tubes.
Generation of whole cell extracts and conditions for EMSAs have been previously described (16, 24). For supershift experiments, Abs were preincubated with 5 μg of protein extract for 10 min at room temperature. Probe was added together with poly(dI-dC), and after an additional incubation for 10 min at room temperature, samples were loaded onto prerun gels. For all whole cell or nuclear extracts that were analyzed for their κB binding activities, parallel EMSA reactions with an octamer probe were performed to control for integrity and concentration of the extracts. For in vitro inhibition of NF-κB/Rel complexes, wheat-germ-translated IκBα and IκBβ proteins were produced (Promega, Madison, WI). In vitro translated proteins were diluted 1/3, 1/9, and 1/27 in buffer containing 25 mM HEPES (pH 7.7), 100 mM KCl, 1 mM EDTA (pH 8.0), and 20% glycerol, supplemented with 1 mM DTT and 1 mM PMSF. One microliter of each dilution was preincubated with 1 μg of protein extract for 10 min at room temperature. After addition of probe and nonspecific competitor, reactions were incubated for 10 min at room temperature and then loaded onto prerun gels.
Western blot analyses were conducted using 40 μg of protein extract. Proteins were blotted onto polyvinyldifluoride membrane, and the membrane was blocked with 7.5% dry milk including 0.2% Tween-20. Subsequent washes were performed in PBS/0.2% Tween-20. Proteins detected by the primary Ab were visualized by the use of an enhanced chemiluminescence assay (Amersham (Arlington Heights, IL) or Boehringer Mannheim (Mannheim, Germany)). Abs used for supershifts, Western immunoblots, and immunoprecipitations were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), except for the p50-specific Ab, which was a gift from Dr. Rodrigo Bravo, Bristol Myers Squibb (Princeton, NJ).
For coimmunoprecipitations, 2 × 107 pre-B cells were washed once with PBS and lysed at 4°C for 30 min under rotation with 1 ml of lysis buffer (50 mM NaF, 250 mM NaCl, 5 mM MgCl2, 1 mM EGTA, 0.5 mM EDTA, 0.001% sodium azide, 0.1% Triton X-100, 20% glycerol, 50 mM Tris (pH 8.0), 1 mM PMSF, and 10 μg/ml leupeptin). After microcentrifugation (12,000 × g, 15 min, 4°C) supernatants were transfered to fresh tubes and incubated with 5 μg of Abs (Santa Cruz) in the presence or the absence of the corresponding peptides (4 μg) for 2 h on ice. Equilibrated protein A-agarose beads (50 μl; Pierce Chemical Co., Rockford, IL) were added and incubated for an additional 1 h. The beads were collected and washed three times with lysis buffer before analysis on 12% SDS-PAGE.
Primary pre-B cells contain low levels of nuclear NF-κB
We analyzed primary pre-B cell lines, which grow in the presence of IL-7 on bone marrow-derived stromal cells, for the presence of active NF-κB. Cells that can be expanded by this method represent very early, committed stem cells of the B lineage (29). Virtually all cells in these cultures express PB76, an early B-lineage marker. Both the κ and the λ light chain genes are in germ-line configuration, but all the cells show DH-JH rearrangements on either one or both alleles. When nuclear extracts from these cells were compared with either an Abelson virus-transformed pre-B cell line (PD31) or a pre-B cell line transformed by a retrovirus bearing an activated Ha-ras oncogene (HAFTL), the same low levels of constitutive nuclear NF-κB complexes were observed (Fig. 1,A). Three main complexes were distinguishable. The amounts of detectable complexes were drastically lower than those in typical mature B cell lines or in B220-sorted splenic B cells (Fig. 1 B and data not shown). The observed complexes were specific for the κB motif, as they could be efficiently competed for by an excess of wild-type, but not mutant, κB binding sites (data not shown).
Constitutively active NF-κB complexes in splenic B cells have previously been identified to be predominantly p50/RelB heterodimers (16, 17, 24). We therefore analyzed the identity of the weak constitutive complexes in pre-B cells using specific Abs. Whereas the RelA-specific Ab removed the upper complex, a slightly faster migrating complex reacted with the RelB-specific Ab. The lowest complex was completely abolished by anti-p50 Abs, which also diminished the other two complexes (Fig. 1 C). Although no clear supershift was observed with the c-Rel-specific Abs, they do remove a complex migrating between the RelA- and RelB-specific complexes. Given that a complex of similar mobility remains after supershifting with the p50-specific Ab, this suggests that c-Rel might be predominantly complexed with RelA as RelA/c-Rel heterodimer in these pre-B cells. Identical results were obtained when the constitutively active complexes from Abelson virus-transformed PD31 cells were characterized (data not shown). In summary, these results demonstrate that both primary and transformed pre-B cells contain low levels of constitutive NF-κB binding complexes, consisting of the various Rel family members.
Differentiation-associated nuclear translocation of NF-κB
We next wanted to determine whether nuclear NF-κB levels would be affected upon differentiation of the pre-B cells. Upon removal of IL-7, cells stop proliferation, and some of the cells differentiate to surface IgM-positive, immature B cells within 72 h (29). Strong induction of NF-κB binding activity was observed upon IL-7 withdrawal (Fig. 2,A). In contrast to conventional NF-κB induction by a multitude of exogenous stimuli, which is typically a very rapid event taking only minutes, NF-κB induction following IL-7 withdrawal was slow. No significant induction was detectable at early time points, but induction could be seen at 18 h of IL-7 withdrawal and continued to increase until 48 h (Fig. 2 A).
Differentiation of the pre-B to B cells is accompanied by completion of rearrangement of the Ig heavy chain locus (VH to DJH). Inherently, this process only leads to a fraction of correct coding joints (one-third for every recombination event) and consequently a significant proportion of unsuccessful rearrangements. After 72 h of differentiation only about 10% of the population has reached the immature surface positive (sIgM+) B cell stage. Many cells remain sIg negative, and a large number of cells die by apoptosis (29). To test whether the observed increase in NF-κB is seen in the sIgM+ population, these cells were enriched by magnetic bead separation. This procedure resulted in a population containing about 70–80% sIgM+ cells. The identical increase in NF-κB binding activity was detected in this enriched cell population, suggesting that the increase in NF-κB activity accompanies the differentiation step (Fig. 2 B).
Specific Abs were used to analyze which NF-κB/Rel family members were induced under these conditions. The majority of the inducible κB binding activity reacted with a RelA-specific Ab, suggesting that p50/RelA represented the main inducible complex. In addition, some c-Rel-containing complexes, most likely RelA/c-Rel heterodimers, were detected in the induced DNA binding complexes. In contrast, the amount of RelB-containing complexes remained low and unaltered during this differentiation process (Fig. 2 C).
An important issue to resolve was whether the observed increase in NF-κB activity was associated with the differentiation program or merely some sort of stress response due to growth factor deprivation. We first analyzed NF-κB induction in a pre-B cell clone expressing the bcl-2-transgene. These cells show a similar differentiation as normal cells upon IL-7 withdrawal, but they are protected from apoptosis (38). A similar level of NF-κB induction as that with normal cells was obtained, albeit with a slightly retarded time curve (Fig. 3,A). The reasons for this delay are presently unclear. We then analyzed pre-B cell lines derived from rag-2-deficient (Rag2T) mice. These cells have been demonstrated to show some signs of differentiation upon IL-7 withdrawal, but due to the lack of DNA rearrangements, these cells cannot differentiate to sIgM+ immature B cells (39). Upon IL-7 withdrawal, only a very weak induction of active NF-κB at 24 h was observed (Fig. 3,A). The same result was obtained when Rag2T pre-B cells expressing the bcl-2 transgene were used. As expected, differentiation by IL-7 withdrawal resulted in induction of κ-RNA levels in wild-type and bcl-2-transgenic cells, whereas no increase was seen for the rag-2-deficient cells (Fig. 3 B). These results suggest that strong activation of NF-κB specifically occurs in pre-B cells that have the ability to differentiate to the sIgM+ stage.
The failure to induce NF-κB in the rag-2−/− cells could be due to the fact that these cells lack expression of NF-κB/Rel proteins. We therefore analyzed extracts from the various pre-B cell lines by Western immunoblotting with Abs specific for different NF-κB/Rel family members. All cell lines contained comparable levels of RelA, RelB, and c-Rel proteins (Fig. 3,C). Similarly, IκBα levels were also indistinguishable in the various cell lines, suggesting that the rag-2 deficiency did not directly imbalance the level of NF-κB/IκB protein expression. Furthermore, the NF-κB/Rel proteins in these cell lines were all functional and inducible. Hydroxyl radicals have been shown to be involved in most pathways leading to NF-κB induction. Therefore, NF-κB induction can be achieved in most cell types by treating them with H2O2 (40) This experiment revealed that both rag-2-positive and negative cell lines all contained comparable levels of inducible NF-κB proteins (Fig. 3 D).
IκBβ degradation correlates with NF-κB induction
The kinetics of NF-κB induction in this B cell differentiation system are clearly different from those of most other NF-κB inductions. In most cases, stimulation of cells results in a very rapid, typically transient induction of p50/RelA (3, 4). In contrast, the induction that we detected in B cells showed very slow kinetics, and the induced complexes did not disappear within the time period analyzed. To determine whether this might be due to a consistent overproduction of RelA or permanent down-regulation of IκBα, we analyzed NF-κB/Rel protein levels by Western immunoblots. Interestingly, neither the RelA nor the IκBα protein levels showed any alterations in any of the cell lines tested (Fig. 4 A). In contrast, both RelB and c-Rel protein levels clearly increased in the wild-type and bcl-2 transgenic cells, but not in either of the two rag-2-deficient cell lines. Interestingly, induction of RelB accumulation is again delayed in the bcl-2 transgenic pre-B cells similar to the appearance of nuclear p50/RelA.
Increased levels of RelB and c-Rel proteins are typical features of mature resting B cells. We therefore compared absolute levels of NF-κB/Rel proteins in undifferentiated pre-B cells, immature B cells generated by IL-7 withdrawal, and B220 sorted primary B cells from spleens. Whereas comparable amounts of RelA were observed at all stages, immature and mature B cells contained significantly increased levels of both RelB and c-Rel (Fig. 4,B, left panel). In fact the amounts of c-Rel and RelB protein present in the cells differentiated by IL-7 removal for 72 h were almost identical with the amounts detected in the mature splenic B cells. We also analyzed the amounts of the different Rel proteins in the sIgM+-selected fraction and observed an identical increase, suggesting that these induced expression levels accompany pre-B to immature B cell differentiation (Fig. 4 B, right panel).
The results from the EMSA experiments had demonstrated that p50/RelA- and, to a lesser extent, c-Rel-containing complexes appear in the nucleus as a consequence of B cell differentiation, whereas no significant increase in p50/RelB can be seen (Fig. 2). Given the strong induction of RelB protein expression observed, these findings suggest that the RelB protein accumulates preferentially in the cytoplasm. We therefore analyzed nuclear and cytoplasmic fractions of pre-B cells and cells differentiated for 3 days by IL-7 withdrawal for the presence of the different Rel proteins. Evidently, there is an increase in the amounts of nuclear p50, RelA, and also c-Rel. In contrast, the newly synthesized RelB protein as well as most of c-Rel are primarily retained in the cytoplasm (Fig. 4 C). Interestingly, although the induction of nuclear DNA-binding NF-κB activity increases significantly during this differentiation process, only a small amount of the cytoplasmic NF-κB pool is mobilized. A similar result was obtained previously when the transformed pre-B cell line 70Z/3 was differentiated to the immature B cell stage by LPS treatment. This also resulted in a strong induction of nuclear NF-κB DNA binding activity of RelA and c-Rel complexes without showing a reduction of the cytoplasmic RelA pool (41). The amounts of cytoplasmic p105/NFKB1, the precursor for p50, did not change during the 3-day differentiation period (data not shown).
The apparent lack of IκBα degradation during B cell differentiation suggested that this NF-κB induction might proceed via an alternate pathway compared with typical NF-κB activation. We therefore investigated whether IκBβ might be involved. Pre-B cells, immature B cells, and B220-positive splenic B cells were analyzed for their IκBα and IκBβ expression levels by Western immunoblot. Whereas IκBα levels were virtually identical at all stages, steady state levels of IκBβ were clearly reduced after 3 days of differentiation in the absence of IL-7 (Fig. 5,A, left panel). This reduction was also seen in the sIgM+-selected fraction (Fig. 5,A, right panel). The slow induction of nuclear p50/RelA observed in this differentiation system suggested that IκBβ degradation should also proceed gradually. We therefore investigated the kinetics of IκBβ degradation by analyzing protein levels at different time points after IL-7 withdrawal. IκBβ protein levels were reduced about twofold after 24 h and continued to decrease to about 15 to 25% within 72 h (Fig. 5 B). The time course of IκBβ loss was again delayed in the bcl-2 transgenic pre-B cells (data not shown), supporting the conclusion that IκBβ degradation is involved in the observed induction of NF-κB during B cell differentiation.
IκBβ is a selective inhibitor of RelA-containing complexes
Why does degradation of IκBβ lead to a rather specific accumulation of nuclear RelA, when at the same time the absolute level of RelB significantly increases in the cytoplasm? A potential explanation could be that IκBβ is a more selective inhibitor than IκBα and primarily targets RelA- and c-Rel-containing complexes. Previous interaction studies had shown that IκBβ is associated with both RelA and c-Rel in B cell lines (7). To address whether IκBβ is associated with all Rel proteins in the primary pre-B cells, we performed coimmunoprecipitations with subsequent immunoblotting of the immune complexes. IκBα- and IκBβ-containing complexes were immunoprecipitated and then analyzed by immunoblots with RelA, RelB, or c-Rel-specific Abs. This analysis revealed that RelA and c-Rel are associated with both IκBα and IκBβ inhibitors. In contrast, RelB was only found associated with IκBα, not IκBβ (Fig. 6 A). The specificity of the immunoprecipitation reaction was controlled by inclusion of an excess of antigenic peptide used to generate the IκBα- and IκBβ-specific Abs. This result suggests that IκBβ does not function as an inhibitor of RelB complexes in pre-B cells. Therefore, increased cytoplasmic RelB complexes must be associated with a different type of inhibitor protein.
These results pose the question of whether IκBβ is as such incapable of interacting with RelB or whether the pre B cell environment is responsible for this observed differential interaction. We therefore performed in vitro inhibition experiments. The different p50/Rel heterodimer complexes were generated by transfection of the respective expression vectors in COS cells and challenged with in vitro translated IκBα and IκBβ proteins. In these experiments, IκBα was an efficient inhibitor of complexes containing all three types of Rel proteins (Fig. 6 B). In contrast, IκBβ specifically inhibited p50/RelA complexes, but no inhibition of RelB or c-Rel complexes was detectable. The failure to inhibit p50/RelB was consistent with the previous result showing a lack of association between IκBβ and RelB. The observation that p50/c-Rel complexes likewise were not inhibited by IκBβ was surprising, however. Association of IκBβ with p50/RelA without inhibiting DNA binding was previously demonstrated for the nonphosphorylated form of IκBβ (42). However, in the experiments described here, the identical IκBβ protein efficiently inhibits p50/RelA but does not target p50/c-Rel.
To exclude the possibility that this might be due to incomplete modification of the IκBβ protein in the in vitro translation experiment, we performed transient cotransfections. A κB-dependent reporter construct was cotransfected with expression vectors for p50 and the different Rel family members in the absence or the presence of increasing amounts of IκBα or IκBβ expression vectors. These results convincingly demonstrate that IκBα is an efficient inhibitor of the transcriptional activity of all three Rel protein complexes (Fig. 6,C). In contrast and consistent with the in vitro inhibition experiments, IκBβ shows a strong selectivity for inhibiting RelA-mediated transcription, but is rather inefficient at inhibiting the transcriptional activity of p50/RelB or p50/c-Rel. Whereas p50/RelA-mediated transcription is reduced about 100-fold, p50/RelB- and p50/c-Rel-dependent transcription are only reduced by 2.5- and 6-fold, respectively. A potential explanation for the observed interaction of IκBβ with c-Rel in vivo yet the failure to efficiently inhibit c-Rel both in vitro as well as upon cotransfection in vivo could be that the observed association of IκBβ with c-Rel in extracts involves RelA/c-Rel heterodimers. It has been demonstrated by a variety of groups that RelA and c-Rel efficiently heterodimerize in vivo and are responsible for transcriptional activation of a subset of NF-κB target genes (43, 44). We therefore analyzed the ability of IκBβ to inhibit transcription driven by cotransfection of RelA and c-Rel expression vectors. Evidently, these heterodimers are subject to IκBβ inhibition (Fig. 6 C).
Here we show that primary pre-B cells, like their transformed counterparts, contain low levels of constitutively active NF-κB. In both cases these complexes represent predominantly p50/RelA and p50/RelB heterodimers. Upon removal of the growth-supporting factor IL-7, these pre-B cells differentiate to surface IgM-positive, immature B cells (29). In the process of this differentiation, a slow increase in NF-κB was observed. This induction is mainly due to activation of p50/RelA and, to a lesser extent, c-Rel-containing heterodimers, whereas p50/RelB complexes remained virtually unaffected. The observation that p50/RelA heterodimers represent the predominant species upon primary B cell maturation is consistent with the earlier report of Singh and colleagues (26). They had analyzed NF-κB proteins in a different primary pre-B cell culture system (Whitlock-Witte cultures). This culture system is more heterogeneous than the one employed here, as in the Whitlock-Witte system pre-B cells continuously differentiate, and a mixed population of pre-B and mature B cells is always present. The strong constitutive signal observed in that system therefore most likely was derived from the more mature B cells present in the population.
It had been suggested that the half-life of the IκBα protein is decreased in mature B and plasma cells compared with that in pre-B cells and that this alteration in the IκBα turnover rate might be important for the increased levels of constitutively active NF-κB (21, 25). Although we have not addressed this question in detail in our system, our results are consistent with this interpretation. IκBα is one of the target genes of NF-κB, and when steady state levels of mRNA levels for IκBα were analyzed, a clear increase was observed (data not shown). In contrast, the amounts of stable IκBα protein remained virtually unaffected during the differentiation. This finding suggests an increase in the turnover rate of the IκBα protein. We could, however, identify a second mechanism involved in the induction of NF-κB during B cell development. IκBβ levels significantly decrease in the process of differentiation and the time course of IκBβ loss mirrors the induction of DNA-binding NF-κB complexes. We note, however, that at early time points (18–24 h), IκBβ levels are only reduced 2-fold, whereas DNA-binding NF-κB is induced about 10- to 15-fold. Our experiments do not distinguish between an active degradation of IκBβ and a reduced synthesis of IκBβ over the period of B cell differentiation. However, the net result is the obvious reduction in steady state levels of IκBβ protein. The conclusion that specific IκBβ loss contributes to the observed activation of predominantly p50/RelA is additionally supported by the inhibition of the specificity of IκBβ for RelA-containing complexes. Moreover, when fibroblasts from RelA-deficient mice were analyzed for IκB levels, a significant down-modulation of IκBβ, but not IκBα, levels was observed, again in line with a specific association between RelA and IκBβ (45).
A different role for IκBβ in regulating the induction of NF-κB in B cells was suggested recently. Upon long-lasting induction of 70Z/3 pre-B cells with LPS, it was found that reappearing IκBβ is predominantly unphosphorylated (42). This unphosphorylated IκBβ can associate with p50/RelA, but does not interfere with nuclear translocation or with DNA binding. In fact, the authors suggested a protective function for the NF-κB/IκBβ heterotrimers, in that they can escape inhibition by IκBα. A role of unphosphorylated IκBβ in the normal pre-B to B cell differentiation process remains to be investigated.
Our observation that IκBβ is a much more selective inhibitor of NF-κB proteins than IκBα has several important physiologic consequences. In combination with the apparently distinct degradation pathways for IκBα and IκBβ, which were obvious in our studies as well as in the previous report (7), a fine-tuned regulation of specific NF-κB activities can be achieved. To date, not all the details of the IκBβ degradation pathway have been elucidated. For IκBα it is known that the various inducing signals result in the phosphorylation of IκBα on two serine residues in the N-terminus, Ser32 and Ser36 (46, 47). This phosphorylation apparently targets IκBα-P for the ubiquitin conjugation enzymes, and polyubiquitinated IκBα is then rapidly degraded by the proteasome (48, 49, 50). IκBβ also contains conserved phosphorylation sites in the N-terminus, which have been shown to be important for inducible phosphorylation and degradation (51). Recent data suggest that IκBα and IκBβ are targeted by the same IκB-kinase (IKKα/CHUK); the efficiency of phosphorylation of IκBβ in vitro was reduced compared with that of IκBα, however (52). Clearly, there must be specific intracellular mediators that distinguish between IκBα and IκBβ and therefore lead to specific degradation of either one or the other. The signaling cascade that ultimately triggers IκBβ degradation is largely unknown. It is unlikely that IL-7-mediated signals are directly involved in this scenario. Signaling via IL-7 and IL-7R would have to provide a negative signal to the cells preventing NF-κB activation. However, as removal of IL-7 from rag-2-deficient pre-B cells did not result in a comparable NF-κB induction, arguing for a more complex regulation. It is more likely that in the process of differentiation the cells activate an endogenous program that leads to IκBβ destabilization and therefore to preferential p50/RelA induction.
A critical role for NF-κB induction during B cell differentiation was recently demonstrated in an independent set of experiments. When different transformed pre-B cell lines were stably transfected with a dominant negative version of IκBα (lacking the N-terminus needed for inducible degradation), rearrangement and transcription of the Igκ locus were blocked (53). Given the broad inhibition spectrum of IκBα, induction of all NF-κB complexes was inhibited in these cells. Therefore, no information could be derived from these experiments as to which NF-κB/Rel family members are induced and which IκB proteins are degraded under physiologic conditions.
We were surprised to find that IκBβ efficiently associates with c-Rel complexes, yet when tested for direct inhibition of DNA binding and transcriptional activity of p50/c-Rel heterodimers, it is apparently quite inefficient. A reduced inhibition level for p50/c-Rel heterodimers compared with p50/RelA upon COS cell cotransfection was noted in the original paper describing the cloning of IκBβ (7). A potential explanation for the discrepancy between association and inhibition could be the following hypothesis. The observed association between IκBβ and c-Rel in vivo could be specific for RelA/c-Rel heterodimers, c-Rel homodimers, or other heterodimers distinct from p50/c-Rel. In support of this hypothesis we showed that transcriptional activation mediated by RelA/c-Rel heterodimers could be efficiently inhibited by IκBβ. Furthermore, when we performed coprecipitation experiments from pre-B cells with RelA-specific Abs, we always observed a large amount of associated c-Rel and vice versa (data not shown). This suggests that pre-B cells contain significant quantities of these RelA/c-Rel heterodimer complexes, which could explain the observed IκBβ association. Additionally, it was shown recently that phosphorylated IκBβ can efficiently inhibit DNA binding of c-Rel homodimers, both in vitro and in vivo (54). Again, these authors noted obvious differences between inhibition of RelA (RelA homodimers) vs c-Rel homodimers by IκBβ, in line with the observations described here. Consistent with this result, we found that IκBβ was more efficient in reducing c-Rel-driven transcription compared with RelB-dependent activation (Fig. 6 C). This could be due to the fact that the activity that we scored there was derived from both p50/c-Rel heterodimers as well as c-Rel homodimers. The latter of the complexes should be inhibited by IκBβ (54). A specific association between RelA and IκBβ in pre-B cells was recently also suggested by Whiteside and colleagues, who noted that >90% of the RelA protein was complexed to IκBβ in 70Z/3 cells (9).
Our finding that RelB protein levels increase during B cell differentiation accompanying the increased levels of active NF-κB together with additional evidence suggest that relb, like c-rel, is also a target gene of NF-κB. Firstly, relb was originally cloned as an immediate early gene induced by serum in starved fibroblasts (55). Secondly, we have recently characterized the defect in a pan-NF-κB-deficient murine plasma cell line (56, 57). We could show that S107 cells, which completely lack nuclear κB binding proteins, contain cytosolic RelA and c-Rel proteins that cannot be induced by any of the known treatments inducing NF-κB. The primary defect in these cells therefore seems to be located in the pathway leading to NF-κB induction. Interestingly, RelB is not at all expressed in these cells, consistent with an essential role of NF-κB for RelB expression. Finally, the induction kinetic of RelB expression during pre-B cell differentiation closely follows the induction of NF-κB (p50/RelA) in our system, again suggesting that NF-κB is a positive regulator of RelB expression.
The primary B cell differentiation system does not recapitulate the complete B cell maturation. The cells generated in this system typically do not express IgD on the surface and were shown to have a much reduced ability to respond to mitogenic stimuli (30). In line with this observation, there is a clear difference between mature primary B cells and the cells generated in this system with respect to constitutively active NF-κB. Whereas both cell types express large quantities of the RelB and c-Rel proteins, the predominant DNA binding species is p50/RelB in the case of mature B cells and p50/RelA in the immature B cells generated by in vitro differentiation (this study and Refs. 16 and 17). It is unclear how the RelB complexes are retained in the cytoplasm of the immature B cells. RelB could lack the specific modification that allows it to escape IκBα-inhibition as suggested previously (24). Alternatively, another IκB-protein, such as the recently identified IκBε, might be involved in retaining RelB.
We thank J. Hess, P. Pfisterer, and S. Zwilling for many helpful comments and critical reading of the manuscript, Y. Cully and S. Pfränger for help with the artwork, and J. Rami for help preparing the manuscript.
This work was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG Wi 789/2-1; to T.W.). The Basel Institute for Immunology was founded and is supported by F. Hoffmann-La Roche Co. (Basel, Switzerland).
Abbreviations used in this paper: NF-κB, nuclear factor-κB; EMSA, electrophoretic mobility shift assay; sIg, surface Ig.