The Ability of CD40L, but Not Lipopolysaccharide, To Initiate Immunoglobulin Switching to Immunoglobulin G1 Is Explained by Differential Induction of NF-κB/Rel Proteins

Abstract

Antibodies of the immunoglobulin G1 class are induced in mice by T-cell-dependent antigens but not by lipopolysaccharide (LPS). CD40 engagement contributes to this preferential isotype production by activating NF-κB/Rel to induce germ line γ1 transcripts, which are essential for class switch recombination. Although LPS also activates NF-κB, it poorly induces germ line γ1 transcripts. Western blot analyses show that CD40 ligand (CD40L) induces all NF-κB/Rel proteins, whereas LPS activates predominantly p50 and c-Rel. Electrophoretic mobility shift assays show that in CD40L-treated cells, p50-RelA and p50-RelB dimers are the major NF-κB complexes binding to the germ line γ1 promoter, whereas in LPS-treated cells, p50–c-Rel and p50-p50 dimers are the major binding complexes. Transfection of expression plasmids for NF-κB/Rel fusion proteins (forced dimers) indicates that p50-RelA and p50-RelB dimers activate the germ line γ1 promoter and that p50–c-Rel and p50-p50 dimers inhibit this activation by competitively binding to the promoter without activating the promoter. Therefore, germ line γ1 transcription depends on the composition of NF-κB/Rel proteins.

After activation by immunization or infection, naive resting B cells expressing IgM and IgD switch to expression of IgG, IgE, and IgA isotypes. Isotype, or class, switching is mediated by a DNA recombination event called class switch recombination. Recombination occurs between two switch regions, one located 5′ to the Cμ gene and the other located 5′ to one of the downstream heavy chain constant region (C_H) genes. Class switching does not alter the antigen specificity of the antibody isotype but does alter its effector function, e.g., the ability to bind to complement, to transcytose across epithelial cells, or to mediate an allergic reaction (41, 44).

Cytokines, B-cell mitogens, and the nature of antigen determine the choice of isotype. For example, the IgG1 and IgG3/IgG2b antibody classes are preferentially induced in mice by T-cell-dependent and -independent antigens, e.g., proteins and bacterial LPS, respectively (31, 40).

An initiating event in class switching is the induction of transcription of the unrearranged, or GL, C_H gene to which the cell will switch. Transcription from the GL C_H gene initiates at exon I, upstream of the switch region of each C_H gene (44). Subsequent RNA splicing to produce mature GL transcripts, also called switch transcripts, appears to be required for switch recombination (26). The effect of targeted disruption of the promoter and exon I of GL C_H gene transcripts provides solid evidence for the requirement for GL transcripts in class switching (4, 20, 54).

Since GL transcripts direct switch recombination, an understanding of the mechanisms of regulation of GL transcripts is necessary for understanding the regulation of class switching. Expression of GL transcripts is regulated at the transcriptional level by cytokines, such as IL-4, gamma interferon, and transforming growth factor β, and by B-cell activators, such as LPS, CD40L, and stimuli that induce signaling through surface Ig (19, 27, 37, 44, 45, 48, 53). For example, IL-4 and CD40L induce GL γ1 transcripts, whereas LPS, in the absence of IL-4, induces GL γ2b and γ3 transcripts (27, 37, 48).

We previously reported that CD40L, but not LPS, induces the activity of the GL γ1 promoter in a reporter plasmid (25). This induction by CD40L is mediated by activating NF-κB/Rel proteins to bind to three tandem κB sites located in the CD40RR of the GL γ1 promoter, which is located just downstream of an IL-4-responsive Stat6 binding site (3). Overexpression of NF-κB/p50 together with RelA or RelB is sufficient to transactivate the GL γ1 promoter or the CD40RR inserted upstream of a minimal c-fos promoter, whereas overexpression of NF-κB/p50 together with c-Rel fails to induce even though this heterodimer binds to the GL γ1 promoter. These observations suggest that NF-κB/Rel proteins are important for initiating class switching to IgG1 in response to T-dependent antigens.

Although LPS is an activator of NF-κB, LPS is a poor inducer of GL γ1 transcripts. In this study, we have addressed the reason for the differential effects of LPS and CD40L on induction of GL γ1 promoter and transcripts. We have tested the hypothesis that GL γ1 transcription depends on the composition of the induced NF-κB/Rel dimers.

MATERIALS AND METHODS

Abbreviations.

Abbreviations used in this paper are as follows: Ig, immunoglobulin; GL, germ line; CD40L, CD40 ligand; CD40RR, CD40 responsive region; LPS, lipopolysaccharide; IL, interleukin; PDTC, pyrrolidine dithiocarbamate; TLCK, Nα-p-tosyl-l-lysine chloromethyl ketone; DMSO, dimethyl sulfoxide; CHX, cycloheximide; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; CAT, chloramphenicol acetyltransferase; RT, reverse transcription; EMSA, electrophoretic mobility shift assay(s); GM-CSF, granulocyte-macrophage colony-stimulating factor; G-CSF, granulocyte colony-stimulating factor; TNF-α, tumor necrosis factor alpha.

Splenic B cells and cell lines.

Splenic B cells from BALB/c mice were prepared by T-cell depletion by adding monoclonal antibodies for Thy1 (J1j.10), CD4 (GK1.5), and CD8 (3.168.3), followed by addition of anti-rat κ chain monoclonal antibody (MAR 18.5) and guinea pig complement. Elutriation was used to obtain small resting B cells. The 1B4.B6 cell line was obtained by immortalizing LPS-stimulated BALB/cByJ splenic B cells by in vitro transfection with the J-2 retrovirus expressing avian v-raf and v-myc, as described previously (6, 32). 1B4.B6 cells are CD45(B220), IgM, and IgD positive (data not shown). 1B4.B6, M12.4.1 (15), and splenic B cells were maintained in a 5% CO₂ incubator in RPMI 1640 medium with 10% fetal bovine serum, 50 mM 2-mercaptoethanol, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 2 mM glutamine, 100 U of penicillin per ml, 100 μg of streptomycin per ml, and 0.1 mg of kanamycin sulfate (GIBCO, Grand Island, N.Y.) per ml. HEK 293 cells, used for overexpression of p50-RelA, p50–c-Rel, p50-RelB, and p50-p50 fusion proteins (see below), were cultured in Dulbecco modified Eagle medium with 10% fetal calf serum.

Soluble CD40L-CD8α fusion protein, recombinant IL-4, LPS, NF-κB inhibitors, and okadaic acid.

Cell culture supernatant containing soluble CD40L-CD8α fusion protein was prepared from J558L mouse myeloma cells stably transfected with the CD40L-CD8α fusion gene as described previously (24, 25). Cell culture supernatant from untransfected J558L cells was prepared as a control. Recombinant mouse IL-4 was a gift of W. E. Paul, National Institutes of Health. LPS (Escherichia coli serotype O55:B5), NF-κB inhibitors PDTC and TLCK, and an NF-κB activator (okadaic acid) were purchased from Sigma. TLCK and okadaic acid were dissolved in DMSO at 75 mM and at 125 μg/ml, respectively, and were diluted 1,000-fold when added to cells. LPS and PDTC were dissolved in RPMI medium.

RNA isolation.

Total RNA was prepared from 1B4.B6 and splenic B cells by the hot phenol method (33).

Semiquantitative RT-PCR. (i) RT.

RT was performed with the following specific primers: 5′-CTGAGCTGCTCAGAGTGTA-3′ (positions 300 to 282, GenBank no. V00793) (18) for GL γ1 transcripts and 5′-TCACAAACATGGGGGCATC-3′ (positions 437 to 419, GenBank no. M32599) (35) for GAPDH transcripts. These two specific primers were mixed with 4 μg of total RNA from splenic B cells or 30 μg of total RNA from 1B4.B6 cells. The mixture was heated at 65°C for 10 min and cooled down to room temperature. To the mixture were added the following reagents: 6 μl of 5× RT buffer (containing 250 mM Tris-HCl [pH 8.3], 375 mM KCl, 50 mM dithiothreitol, and 15 mM MgCl₂), 2.5 μl of 2.5 mM deoxynucleoside triphosphate, 10 U of RNasin (Promega Corp., Madison, Wis.) and 200 U of Moloney murine leukemia virus reverse transcriptase (Promega). Diethyl pyrocarbonate-treated water was added to obtain a 30-μl volume. The mixture was incubated at 39°C for 1 h to synthesize the cDNAs of GL γ1 and GAPDH transcripts.

(ii) PCR.

Primers used to amplify the GL γ1 cDNA were 5′-ACAGCCTGGTGTCAACTAG-3′ (top strand positions 1772 to 1790, GenBank no. M12389) (28) and the Cγ1-specific RT primer. The primers used to amplify the GAPDH cDNA were 5′-CAAATTCAACGGCACAGTC-3′ (positions 202 to 220, GenBank no. M32599) (35) and the GAPDH-specific RT primer. For PCR amplification of cDNAs from 1B4.B6 cells, the 50-μl reaction volume contained 75 pmol of each primer, 5 μl of RT product, 1× PCR buffer (5 mM Tris-HCl [pH 8.3], 42.5 mM KCl, and 0.1% Triton X-100), and MgCl₂ (1.4 mM for GL γ1 RNA and 3 mM for GAPDH RNA). The cDNA of GL γ1 transcripts was amplified for 40 cycles (1.5 min at 94°C, 2 min at 62°C, and 2 min at 72°C), and the cDNA of GAPDH transcripts was amplified for 20 cycles (1.5 min at 94°C, 2 min at 55°C and 2 min at 72°C). Similar PCR conditions were used for PCR amplification of cDNA from splenic B cells, except that 15 μl of RT product and 1× PCR buffer (27.5 mM KCl and 0.1% Triton X-100) were used for amplification of GL γ1 cDNA for 44 cycles and GAPDH cDNA was amplified for 28 cycles. PCR product (30 μl) was loaded onto an 8% polyacrylamide gel, which was subsequently stained with ethidium bromide to visualize the bands. The specificity of the amplified GL γ1 fragment from 1B4.B6 and splenic B cells was determined in the early experiments by digestion with three restriction enzymes and also by DNA sequencing.

Plasmids. (i) Plasmids for expression of p50-RelA, p50–c-Rel, p50-RelB, and p50-p50 fusion proteins.

(a) The eukaryotic expression plasmid p50-RelA(pEF-1α) was generated for expression of a p50-RelA fusion protein. The relA gene was excised from the pcDNA-I vector (39) with NdeI and XbaI digestion, and the p50 gene was excised from pcDNA-I (25) with EcoRV and HindIII digestion. Both gene fragments were cloned into the pEF1α-neo vector (23) between the EcoRI and XbaI sites with the p50 coding region located 5′ to the RelA coding region. A 19-amino-acid linker between the two gene products was obtained by insertion of the HindIII/EcoRV segment from the multiple cloning region of the pcDNA-3 vector. (b) To generate the p50–c-Rel(pEF-1α) plasmid for expression of a p50–c-Rel fusion protein, the c-rel gene in pcDNA-I (25) was excised with NcoI and XbaI and used to replace the relA gene between the NdeI and XbaI sites of the p50-RelA(pEF-1α) plasmid. (c) For generation of the expression plasmid for the p50-RelB fusion protein, the p50 and relB genes were cloned into the pEF1α-neo vector by the following procedure. The HindIII fragment from pcDNA-I containing the p50 coding region (25) was inserted into the HindIII site of the pcDNA-3 vector (Invitrogen, San Diego, Calif.). Then, p50 cDNA was excised from pcDNA-3 with EcoRV and XbaI digestion and cloned into pEF1α-neo vector digested with EcoRI and XbaI. The EcoRI/SacII fragment at the 5′ end of relB cDNA in pcDNA-I (25) was replaced with a PCR fragment produced by using primers 5′-GCTGGAATTCTGCAGATAATGCCGAGTCGCCGCGCT-3′ and 5′-CACTCGTAGCGGAAGCGCAT-3′ and relB cDNA in pcDNA-I (25) as the template. The relB gene was then excised from this plasmid by EcoRI and inserted into pEF1α-neo containing the p50 gene at the EcoRI site. (d) To generate p50(pEF-1α), p50 cDNA, excised from pcDNA-I by EcoRV and XbaI digestion, was inserted into pcDNA-3 digested with EcoRV and XbaI. The p50 cDNA in pcDNA-3 was then excised with EcoRI and XbaI digestion and cloned into pEF1α-neo vector digested with EcoRI and XbaI. To generate p50-p50(pEF-1α), the eukaryotic expression plasmid for the p50-p50 fusion protein, p50 cDNA in p50-RelA(pEF-1α) was excised with EcoRI and inserted into p50(pEF-1α) at the EcoRI site.

All plasmids for expression of NF-κB/Rel fusion protein contain p50 cDNA located upstream of the other NF-κB/Rel gene. The upstream p50 gene is terminated at an internal HindIII site, resulting in a 27-amino-acid deletion at the C terminus of p50 protein and deletion of the stop codon. p50 lacking the C-terminal 27 residues has been found to be able to dimerize with other NF-κB/Rel proteins (13). The amino acid sequence of the linker between two proteins (GTELGSTSNGRQCAGILQI), which was confirmed by DNA sequencing, is the same for all fusion proteins except that the p50-p50 fusion protein contains 3 additional amino acids (SMA) at the C terminus of the linker. The plasmids were transfected into M12.4.1 or HEK 293 cells, and expression of fusion proteins was detected with anti-RelA, anti-c-Rel, or anti-RelB antibody.

(ii) Other plasmids.

The −954WT luciferase reporter gene plasmid contains the GL γ1 promoter fragment from nucleotides −954 to +202 relative to the first RNA initiation site (51). The pCD40FL-F plasmid has the CD40RR from the GL γ1 promoter inserted upstream of a mouse minimal (−71) c-fos promoter segment in a luciferase reporter plasmid (25). The pFosCAT plasmid, containing a minimal (−71) c-fos promoter segment ligated upstream of a CAT reporter gene, was used for internal control of transfection efficiency (14). The previously generated RelA expression plasmid (25) was cotransfected with p50(pEF-1α) plasmid into HEK 293 cells. The ɛ-162Luc plasmid contains the mouse GL ɛ promoter fragment, as described previously (7).

DNA transfections.

Electroporation was performed as described previously for transient transfection of M12.4.1 cells (25). Cells (2 × 10⁷) were washed once with serum-free RPMI medium and were resuspended in 1 ml of serum-free RPMI medium. The cell suspension was mixed with plasmid DNA (a total of about 95 μg, varied to maintain equimolar amounts) and electroporated at 1,250 μF and 750 V/cm. After transfection, cells were left at room temperature for 10 min and then cultured in 10 ml of complete RPMI medium in six-well plates.

The calcium phosphate method (21) was used for expression of NF-κB fusion proteins in HEK 293 cells by transient transfection. Cells (2 × 10⁶) were transfected with 5 μg of plasmid and harvested 48 h after transfection. Nuclear extracts were obtained for EMSA.

Luciferase and CAT assays.

Luciferase (5) and CAT (29) assays were performed and analyzed as described previously (25).

Cytoplasmic and nuclear extracts.

The modified small-scale method for extraction of cytoplasmic and nuclear proteins was described previously (25, 36).

Western blot analysis.

Two micrograms of nuclear extracts from splenic B cells or 10 μg of nuclear extracts from M12.4.1 or 1B4.B6 cells was fractionated on 10% reducing sodium dodecyl sulfate-polyacrylamide gels and transferred to Immobilon-P membranes (Millipore, Bedford, Mass.) according to the manufacturer’s instructions. The membrane was first incubated with antibody against specific NF-κB/Rel family members in TBS-T buffer (20 mM Tris-HCl [pH 7.6], 137 mM NaCl, and 0.5% Tween 20) in the presence of 5% powdered skim milk overnight. After a wash with TBS-T buffer, the blot was incubated with goat anti-rabbit peroxidase conjugate (Santa Cruz Biotechnology, San Diego, Calif.) for 5 h. The immunoreactive bands were visualized on films by using the ECL system (Amersham Corp.). To remove antibodies so that the blot could be used for further detection of other NF-κB/Rel family members, the blot was incubated with antibody-stripping buffer (62.5 mM Tris-HCl and 100 mM 2-mercaptoethanol) at 50°C for 30 min.

EMSA.

Nuclear extract was combined with 1 μg of annealed poly(dI-dC) · poly(dI-dC) (Sigma) and ³²P-labeled CD40RR fragment in a reaction mixture containing 10% glycerol, 17.5 mM HEPES (pH 7.5), 5 mM KCl, 103 mM NaCl, 0.35 mM EDTA, 0.25 mM EGTA, and 1 mM dithiothreitol. The mixture was incubated at room temperature for 30 min, and samples were loaded onto a 5% nondenaturing polyacrylamide gel (acrylamide/bisacrylamide ratio, 37.5:1) and electrophoresed in 0.5× Tris-borate-EDTA buffer at 150 V. Gels were dried and subjected to autoradiography. For EMSA with antibody supershift, antibodies were premixed with nuclear extracts for 15 min before labeled probe was added. Probe was labeled as previously described (25).

Compared to the reaction conditions reported previously (25), a higher, more physiological salt concentration (total of 126 mM) in the reaction mixture was used in EMSA in this study, as formation of DNA-protein complexes is sensitive to salt concentration. The p50-RelA, p50–c-Rel, and p50-p50 complexes in EMSA with nuclear extracts from M12.4.1 cells were barely detected in our previous study (25), in which 30 mM salt was used, including 5 mM KCl (not 50 mM, as in reference 25, a typographic error), whereas these complexes can be detected under the higher salt conditions, as indicated in Fig. 4A. In addition, under the low-salt conditions previously used, an additional RelB-containing complex was detected and supershifted by anti-RelB but not by other NF-κB antibodies, although it can be competed away by an oligonucleotide with two κB sites. Under the higher salt conditions, all complexes can be supershifted by antibody to p50 (see Fig. 4A; also data not shown).

FIG. 4.

EMSA and densitometry analyses to compare NF-κB activation by LPS and CD40L. (A) Labeled CD40RR probe was incubated with nuclear extracts (NE) from M12.4.1, 1B4.B6, and splenic B cells treated with CD40L or LPS for 12 h to determine binding activity of each NF-κB/Rel dimer. The three major DNA-protein complexes formed are indicated. To supershift or deplete the specific NF-κB/Rel dimers, nuclear extracts were preincubated with antibody against a specific NF-κB/Rel protein, a control serum, or the indicated combinations of antibodies before incubation with labeled CD40RR probe. The total amount of antibody was kept constant at 3 μl by adding control antibody. (B) Densitometry analyses of the NF-κB/Rel complexes from EMSA results shown in panel A. The quantity of binding activity of complex I, containing p50-RelB, was obtained by subtracting the signals at the position of complex I in lanes 8 and 16 from the signals of complex I in lanes 3 and 11, respectively. The quantity of complex III, containing p50-p50, was obtained by subtracting the signals at the position of complex III in lanes 4 and 12 from the signals of complex III in lanes 3 and 11, respectively. The quantity of p50–c-Rel in complex II was obtained by subtracting the signals of complex II in lanes 5 and 13 from the signals of complex II in lanes 3 and 11, respectively. The quantity of p50-RelA in complex II was obtained by subtracting the signals of complex II in lanes 6 and 14 from the signals of complex II in lanes 5 and 13, respectively. The quantity of p50-RelB in complex II was obtained by subtracting the signals at the position of complex II in lanes 7 and 15 from the signals of complex II in lanes 6 and 14, respectively.

Antibodies against NF-κB/Rel and IκB proteins.

Antibodies against p50 (sc-114), c-Rel (sc-071X), RelB (sc-226X), IκBα (sc-371), and IκBβ (sc-945) for Western blot analyses were purchased from Santa Cruz Biotechnology. Antiserum against RelA (34) for Western blot analyses was a gift from N. Rice (National Cancer Institute, Bethesda, Md.). Antibodies to specific NF-κB/Rel family members, including anti-RelA (sc-109X), and antibody against ATF-2 as the control antibody for EMSA were all from Santa Cruz Biotechnology.

RESULTS

GL γ1 transcripts can be induced by CD40L but are only poorly induced by LPS.

Using luciferase reporter assays in M12.4.1 B lymphoma cells, we found previously that CD40L, but not LPS, induces the GL γ1 promoter. The different effects of CD40L and LPS on the GL γ1 promoter are consistent with previous observations that T-dependent antigens, but not T-independent antigens or LPS, induce IgG1 production in mice (31, 40). To determine if CD40L and LPS show a differential ability to induce GL γ1 transcripts in normal B cells, splenic B cells were stimulated with CD40L or LPS for 6 h, and the levels of GL γ1 transcripts were assayed by RT-PCR. Figure 1 shows that CD40L, but not LPS, induces GL γ1 transcripts in splenic B cells. CD40L-induced GL γ1 transcripts persist for at least 18 h, whereas they are still not induced by LPS at this time (data not shown). The results are consistent with a previous report showing that treatment for 18 h with CD40L expressed on Sf9 insect cells induces GL γ1 transcripts in splenic B cells (48).

FIG. 1.

RT-PCR analysis to demonstrate induction of GL γ1 transcripts in splenic B cells. Resting splenic B cells were left untreated or were treated with LPS (50 μg/ml), control supernatant (sup.) (20%), or CD40L supernatant (20%) for 6 h in the presence or absence of the NF-κB inhibitor PDTC (50 μM) or TLCK (75 μM). Expression of GL γ1 transcripts was assayed by RT-PCR with amplification of GAPDH transcripts as the internal control.

To obtain a system that could be more readily manipulated, we examined the effects of CD40L and LPS on GL γ1 transcripts in the B-cell line 1B4.B6. This B-cell line was derived by transformation of LPS-stimulated splenic B cells with the J-2 retrovirus expressing avian v-Raf and v-Myc. This B-cell line has been shown to secrete IgG1 upon stimulation with Th2 cells or with CD40L plus IL-4 (50a). 1B4.B6 cells were treated with CD40L or LPS for various times, and the levels of GL γ1 transcripts were assayed by RT-PCR. As shown in Fig. 2A, CD40L induces GL γ1 transcripts within 2 h and the RNA levels gradually increase for up to 24 h. In contrast, LPS only poorly induces GL γ1 transcripts. Since M12.4.1 cells do not contain GL Cγ1 genes (24a), regulation of expression of GL γ1 transcripts could not be studied in this cell line.

FIG. 2.

RT-PCR analysis of induction of GL γ1 transcripts in 1B4.B6 B cells by LPS, CD40L, and okadaic acid. (A) 1B4.B6 cells were left untreated or were treated with LPS, control supernatant (sup.), or CD40L supernatant, and cells were harvested at the time points indicated for detection of GL γ1 transcripts by RT-PCR. (B) Expression of GL γ1 transcripts was analyzed in 1B4.B6 cells, untreated or treated with control supernatant CD40L supernatant, or IL-4 (1,000 U/ml) for 6 h in the presence or absence of the NF-κB inhibitor PDTC or TLCK. Induction of transcripts by okadaic acid (125 ng/ml), an NF-κB activator, was evaluated at 6 h.

Induction of GL γ1 transcripts by CD40L is mediated by NF-κB activation.

We found previously that induction of the GL γ1 promoter by CD40L in M12.4.1 cells is mediated by activation of NF-κB/Rel proteins that bind three tandem κB sites in the CD40RR (−99 to −43, relative to the first RNA initiation site) of the promoter (25). To determine whether induction of GL γ1 transcripts by CD40L in 1B4.B6 and splenic B cells is also mediated by activation of NF-κB/Rel proteins, we tested the effects of addition of NF-κB inhibitors on induction of the GL γ1 transcripts. Two inhibitors which each inhibit NF-κB activation by different mechanisms were used, the anti-oxidant PDTC and the protease inhibitor TLCK. 1B4.B6 or splenic B cells, pretreated with PDTC at 50 μM or TLCK at 75 μM for 30 min, were incubated with CD40L in the presence of PDTC or TLCK for 6 h. At this time, greater than 90% of PDTC- or TLCK-treated 1B4.B6 cells and TLCK-treated splenic B cells and about 70% of PDTC-treated splenic B cells were viable (data not shown). The levels of GL γ1 transcripts were assayed by RT-PCR. As shown in Fig. 1 and 2B, induction of GL γ1 transcripts by CD40L in splenic B cells and in 1B4.B6 cells is inhibited by PDTC and by TLCK. The inhibition of induction of GL γ1 transcripts by blocking of NF-κB activation appears to be specific to CD40L treatment, because TLCK has no effect on induction of GL γ1 transcripts by IL-4 in 1B4.B6 cells (Fig. 2B).

To determine if the inhibitors did indeed eliminate NF-κB activation by CD40L, an aliquot of the inhibitor-treated 1B4.B6 or splenic B cells used to assay the levels of GL γ1 transcripts were used for preparation of nuclear extracts for EMSA or for Western blot analyses, respectively. We determined previously that all complexes that bind to the CD40RR in EMSA contain NF-κB/Rel proteins (25). The results confirmed that NF-κB DNA binding activity induced by CD40L was abolished by treatment of 1B4.B6 cells with PDTC or TLCK and that NF-κB activation in splenic B cells was inhibited by PDTC and TLCK treatment (data not shown). In conclusion, induction of GL γ1 transcripts by CD40L, but not by IL-4, requires NF-κB activation.

To examine whether NF-κB activation is sufficient to induce GL γ1 transcripts, we tested whether GL γ1 transcripts can be induced by treatment with okadaic acid, which activates NF-κB by stimulating activity of an IκB kinase, resulting in degradation of IκB proteins (8). As expected, okadaic acid treatment for 6 h was found to activate NF-κB binding activity in 1B4.B6 cells (data not shown) and to induce GL γ1 transcripts (Fig. 2B). In addition, we found that the GL γ1 promoter in the −954WT reporter plasmid is induced threefold by okadaic acid in M12.4.1 cells (data not shown). The effect of okadaic acid treatment on GL γ1 treatment in splenic B cells cannot be examined due to induction of cell death. In conclusion, experiments using okadaic acid together with previously reported experiments showing that overexpression of NF-κB/Rel proteins induces GL γ1 promoter activity (25) demonstrate that NF-κB activation is sufficient for induction of GL γ1 transcripts.

NF-κB activation by CD40L and LPS.

Since CD40L-activated NF-κB/Rel proteins induce the GL γ1 promoter, we might expect that LPS, a known NF-κB activator, would also induce the promoter. However, LPS has very little effect on the GL γ1 promoter (25) or on the levels of GL γ1 transcripts (Fig. 1 and 2A), although LPS activates NF-κB/Rel proteins in M12.4.1, 1B4.B6, and splenic B cells (see below). Three possible mechanisms could explain this discrepancy. The first is that CD40L, but not LPS, induces a coactivator protein(s) which is required for promoter activation by NF-κB/Rel proteins. This possibility is not supported by the finding that overexpression of p50-RelA or p50-RelB is sufficient to transactivate the GL γ1 promoter and a reporter gene driven by a minimal c-fos promoter and the CD40RR (pCD40FL-F plasmid) (25). The second possibility is that LPS induces a repressor protein(s) which interacts with NF-κB/Rel proteins and suppresses their transactivation activity. The LPS-induced repression would act at the CD40RR, because LPS cannot induce expression of the pCD40FL-F plasmid. Therefore, the LPS-induced repressor protein(s) must bind to the CD40RR or associate with NF-κB/Rel proteins bound to the CD40RR. However, CD40L and LPS induce similar protein complexes, all of which contain NF-κB/Rel proteins, to bind the CD40RR in EMSA, suggesting that no additional repressor protein is induced by LPS (see Fig. 4A, lanes 3 and 11, and further description below).

The third possibility is that CD40L and LPS might activate different profiles of NF-κB/Rel proteins. We have shown that overexpression of p50 alone or together with c-Rel cannot transactivate the GL γ1 promoter, but overexpression of p50 together with RelA or RelB strongly transactivates the promoter. Therefore, it is possible that LPS induces more p50 and c-Rel proteins and/or fails to induce RelA and RelB proteins. To test this possibility, we compared the patterns and kinetics of NF-κB activation induced by CD40L and LPS. The nuclear level of each NF-κB/Rel protein was determined by Western blot analyses of nuclear extracts from M12.4.1, 1B4.B6, and splenic B cells, and cells from the same culture were used for analyzing induction of promoter activity by a reporter gene assay or induction of GL γ1 RNA by RT-PCR. Two or three independent experiments were performed with nearly identical results for each cell type.

As shown in Fig. 3A and B, CD40L induces greater nuclear accumulation of NF-κB/Rel proteins than does LPS in both splenic B cells and M12.4.1 cells, with a greater difference between CD40L and LPS treatments in splenic B cells and at later time points in M12.4.1 cells. All NF-κB/Rel proteins can be induced by CD40L for at least 12 h in M12.4.1 cells and for at least 24 h in splenic B cells, although RelA levels are slightly reduced at longer time points. In contrast, the induction by LPS of RelA and RelB declines to levels lower than those induced by CD40L at 12 h. However, c-Rel is more persistently induced than RelA and RelB during LPS stimulation.

FIG. 3.

Western blot analyses of NF-κB activation by CD40L and LPS treatment. Nuclear extracts from splenic B cells (A) and from the mouse B-cell lines M12.4.1 (B) and 1B4.B6 (C), left untreated or treated with LPS, control supernatant (sup.), or CD40L supernatant for the indicated times, were analyzed. (D) Western blot analysis with nuclear extracts from M12.4.1 B cells untreated or treated for 6 h with control supernatant, CD40L supernatant, or LPS in the presence or absence of CHX (5 μg/ml). Each Western blot was probed sequentially with four antibodies against different NF-κB/Rel proteins. These experiments were performed two or three times with nearly identical results.

In 1B4.B6 cells, LPS activates NF-κB better than does CD40L and induces significant amounts of RelA and RelB (Fig. 3C). Therefore, poor induction by LPS of GL γ1 transcripts in 1B4.B6 cells is not due to its inability to induce nuclear translocation of the transactivating NF-κB/Rel proteins RelA and RelB. The induction of NF-κB/Rel proteins by CD40L in 1B4.B6 cells occurs with slow kinetics and is still increasing 24 h after addition of CD40L (Fig. 3C). These data are consistent with the gradual increase in GL γ1 transcripts in CD40L-treated 1B4.B6 cells (Fig. 2A). Most importantly, LPS, but not CD40L, stably induces abundant nuclear c-Rel and p50 in this cell line.

Taken together, the Western blot results from the two B-cell lines and splenic B cells indicate that CD40L and LPS induce different patterns of NF-κB activation. Three conclusions can be drawn. (i) LPS induces more c-Rel, and in 1B4.B6 cells also preferentially induces p50, than other NF-κB/Rel proteins, suggesting that p50-p50 and/or p50–c-Rel dimers may be the predominant NF-κB/Rel complexes accumulating in the nucleus in response to LPS treatment. This conclusion is subject to the caveat that the NF-κB antibodies used may have different sensitivities, but data presented in the next section, with EMSA used to quantitate individual NF-κB/Rel proteins, support this conclusion. (ii) CD40L induces greater amounts of RelA and RelB, in comparison to LPS, and also induces p50 and c-Rel. (iii) The kinetics of NF-κB activation show that nuclear translocations of individual NF-κB/Rel proteins are regulated differently by different stimuli.

The p50–c-Rel heterodimer and p50 homodimer are the main NF-κB/Rel dimers bound to the GL γ1 promoter in cells treated with LPS.

To determine if the different pattern of induction of NF-κB/Rel proteins by CD40L in comparison to LPS results in different amounts of individual NF-κB/Rel complexes binding to the GL γ1 promoter, EMSA were performed to examine NF-κB/Rel complexes formed with the CD40RR segment of the GL γ1 promoter and nuclear extracts from M12.4.1, 1B4.B6, and splenic B cells treated with CD40L or LPS for 12 h. Nuclear extracts from CD40L-treated cells form three major complexes with the CD40RR probe, as do nuclear extracts from LPS-treated cells (Fig. 4A, lanes 3 and 11). As indicated in Fig. 4A, complex I contains p50-RelB and complex II consists of p50–c-Rel, p50-RelA, and p50-RelB dimers. Complex II can be entirely supershifted if the amount of anti-p50 antibody is increased to 3 μl (data not shown). Since the CD40RR has three κB sites, it is possible that complex I contains CD40RR DNA fragments with two or three κB sites occupied. Complex III consists of p50-p50 homodimers, as it can be supershifted by anti-p50 but not by antibodies against other NF-κB/Rel family members (Fig. 4A) (25).

To discern the contribution of the individual NF-κB/Rel dimers to the binding activity, we depleted or supershifted the specific NF-κB/Rel dimers by antibodies and used densitometry analyses to determine the amounts of individual NF-κB/Rel dimers binding to the promoter (Fig. 4B). As shown in Fig. 4A, anti-c-Rel antibody eliminates the majority of binding activity of complex II from nuclear extracts of LPS-treated cells (lanes 3 and 5). Addition of anti-c-Rel and anti-RelA to these nuclear extracts almost completely abolishes binding activity of complex II (lane 6). These results suggest that p50–c-Rel, predominantly induced by LPS, is the major NF-κB/Rel dimer in complex II and that some p50-RelA is also induced. Little or no RelB is detected in complex II (lane 6 versus lane 7 and lane 3 versus lane 8), consistent with the poor induction of complex I (lane 3). Also, LPS induces complex III in 1B4.B6 cells. We also examined earlier time points (1 and 6 h) after treatment of 1B4.B6 cells with LPS and again found that p50–c-Rel is the predominant complex (data not shown).

Densitometry analyses of these EMSA (Fig. 4B) show that a greater amount of p50–c-Rel plus p50-p50 binding activity than p50-RelA plus p50-RelB binding activity is detected in nuclear extracts from LPS-treated cells. This conclusion is consistent with the Western blot results in Fig. 3 except for the finding that p50-RelB binding activity is barely detectable in nuclear extracts from LPS-treated cells. This inconsistency suggests that most of the nuclear RelB may be unavailable for binding the CD40RR. A similar finding has recently been reported for RelB induced by anti-Ig treatment of splenic B cells (11).

When we examined the NF-κB/Rel binding activities in nuclear extracts from CD40L-treated cells, it was apparent that anti-c-Rel only partially depletes (M12.4.1 cell extracts) or barely depletes (splenic B and 1B4.B6 cell extracts) the binding activity of complex II (Fig. 4A, lanes 11 and 13). The remaining binding activity of complex II can be greatly reduced by addition of anti-RelA (lane 14). Depletion of c-Rel and RelA reveals the p50-RelB component in complex II, which is completely eliminated by addition of anti-RelB (lane 15). These data indicate that p50-RelA and p50-RelB are the main components in complex II. Therefore, in contrast to LPS, CD40L induces much greater p50-RelB binding activity; this is also indicated by the induction of complex I, which appears to consist mostly of p50-RelB. Densitometry analyses of the EMSA results (Fig. 4B) demonstrate that the combined binding activities of p50-RelA and p50-RelB are greater than those of p50–c-Rel plus p50-p50 dimers in nuclear extracts from cells treated with CD40L.

Generation of p50-p50, p50-RelA, p50–c-Rel and p50-RelB fusion proteins.

The DNA-binding data suggest that the different effects of CD40L and LPS on the expression of GL γ1 transcripts are due to differential induction of NF-κB/Rel dimers binding to the promoter. These NF-κB/Rel dimers might influence the promoter activity by competing with each other for binding to the promoter. To examine this possibility, we tested the effects on the GL γ1 promoter of overexpressing specific NF-κB/Rel dimers. The conventional method for overexpressing NF-κB/Rel dimers is to cotransfect cells with plasmids for individual NF-κB/Rel proteins. However, overexpressed NF-κB/Rel subunits will form various homodimers and heterodimers, as demonstrated by coexpression of p50 and RelA (Fig. 5, lane 8); thus, the effect of a specific NF-κB/Rel dimer cannot be evaluated.

FIG. 5.

EMSA to demonstrate the ability of NF-κB/Rel fusion proteins to bind the CD40RR. EMSA was performed with labeled CD40RR fragment as the probe and 2 μg of nuclear extracts (NE) from HEK 293 cells transiently transfected with empty vector or with expression plasmids for p50-RelA, p50–c-Rel, p50-RelB, and p50-p50 fusion proteins (lanes 2 to 6). Nuclear extracts from cells transfected with p50 plasmid alone (lane 7) or both p50 and RelA plasmids (lane 8) demonstrate formation of multiple NF-κB complexes when two NF-κB/Rel subunits are coexpressed. Nuclear extracts from M12.4.1 cells (3 μg) (lane 9) and 1B4.B6 cells (6 μg) (lane 10) treated with CD40L for 12 h illustrate complexes formed by endogenous NF-κB/Rel dimers.

To overexpress specific NF-κB/Rel dimers, we created plasmids for expressing NF-κB/Rel fusion proteins (p50–c-Rel, p50-RelB, p50-RelA, and p50-p50) by fusing cDNA encoding p50 protein with cDNA encoding p50, RelA, c-Rel, or RelB. To evaluate the expression of the NF-κB/Rel fusion proteins, Western blot analyses were performed with nuclear extracts from M12.4.1 cells stably expressing p50–c-Rel, p50-RelB, p50-RelA, and p50-p50 (Fig. 6). The results are compared with nuclear extracts from HEK 293 cells transiently transfected with the same plasmids. HEK 293 cells do not express nuclear NF-κB proteins, thus allowing a distinction between endogenous and transfected NF-κB proteins. As shown in Fig. 6, intact fusion proteins can be detected in nuclear extracts of M12.4.1 cells and of HEK 293 cells. The fusion proteins, except for p50-p50, can also be detected in M12.4.1 cytoplasmic extracts (data not shown). These results indicate that expressed p50–c-Rel, p50-relB, and p50-RelA can be kept in the cytoplasm, presumably by associating with IκB proteins, and also that they can be translocated into the nucleus. Treatment of the M12.4.1 transfectants with CD40L or LPS induces nuclear accumulation of the NF-κB/Rel fusion proteins (Fig. 6D and data not shown).

FIG. 6.

Western blot analyses demonstrate that NF-κB/Rel fusion proteins can be stably expressed and translocated into the nucleus in M12.4.1 and HEK 293 cells. (A to C) Nuclear extracts were prepared from M12.4.1 cells (stably) or HEK 293 cells (transiently) transfected with the p50–c-Rel, p50-p50, or p50-RelB plasmid. The fusion proteins and endogenous NF-κB proteins in the nuclear extracts were detected by Western blot analyses using anti-c-Rel, anti-p50, and anti-RelB antibodies. (D) M12.4.1 cells, untransfected or stably transfected with the p50-RelA expression plasmid, were left untreated or were treated with LPS, control supernatant (sup.), or CD40L supernatant for 6 h. p50-RelA and RelA in the nuclear extracts were detected by using anti-RelA in the Western blot analysis. Expression of intact p50-RelA in transiently transfected HEK 293 cells was also detected in the nucleus.

To test whether overexpressed fusion proteins can form DNA-protein complexes similar to endogenous NF-κB/Rel dimers, we investigated the ability of p50-p50, p50-RelA, p50–c-Rel, and p50-RelB fusion proteins to bind to the CD40RR. In order to ensure that the binding activity was due to the transfected NF-κB/Rel dimers, NF-κB/Rel fusion proteins overexpressed in HEK 293 cells by transient transfection were examined. As shown by EMSA, overexpressed p50-RelA, p50–c-Rel, p50-RelB, and p50-p50 fusion proteins can bind the CD40RR (Fig. 5, lanes 3 to 6). When the DNA-protein complexes formed by fusion proteins are compared to complexes formed by endogenous dimers in B-cell nuclear extracts (Fig. 5, lanes 9 and 10), it can be seen that p50-RelA, p50–c-Rel, and p50-RelB form complexes which migrate slightly more rapidly than complex II and that the complex formed by the p50-p50 fusion protein migrates clearly faster than complex III. This more rapid migration is probably due to the 27-amino-acid deletion of the C terminus of p50 produced during creation of the fusion proteins. The C-terminal 27 residues of p50 have been documented to be not required for NF-κB dimerization and DNA-binding activity (13). In addition, overexpressed p50-RelB, and to a lesser extent p50-RelA, form an additional slowly migrating complex, presumably corresponding to complex I. p50-RelA also forms a rapidly migrating complex, probably due to protein degradation. Taken together, these results suggest that the fusion proteins have structures similar to that of endogenous NF-κB/Rel dimers.

Overexpression of p50-p50 or p50–c-Rel protein inhibits transactivation activity of p50-RelA and p50-RelB proteins.

To test if p50-RelA, p50-RelB, and p50–c-Rel fusion proteins can transactivate the GL γ1 promoter, the luciferase reporter plasmid containing the −954WT segment was cotransfected with plasmids expressing NF-κB/Rel fusion proteins into M12.4.1 cells and the promoter activity was determined 9 h after transfection. Overexpression of p50-RelA or p50-RelB fusion protein transactivates the promoter activity, whereas overexpression of p50–c-Rel fusion protein poorly induces the promoter (Fig. 7). These data are consistent with previous results showing that p50 and RelA or p50 and RelB activate the promoter whereas p50 and c-Rel or p50 alone do not (25).

FIG. 7.

Reporter gene assays demonstrate that overexpression of p50–c-Rel or p50-p50 fusion protein suppresses the transactivation activity of p50-RelA and p50-RelB fusion proteins. The −954WT luciferase reporter plasmid (20 μg) that contains the GL γ1 promoter and plasmid pFosCat (15 μg) as the transfection control were cotransfected into 2 × 10⁷ M12.4.1 cells with empty vector or different amounts of expression plasmids for NF-κB/Rel subunits or fusion proteins, as indicated (in picomoles). Empty vector was added to equalize the amount of DNA in each transfection. Luciferase activity, representing the promoter activity, was assayed 9 h after transfection and normalized by CAT activity. The fold induction was calculated by the ratio of the luciferase activity from cells transfected with NF-κB/Rel plasmids to the luciferase activity from cells transfected with empty vector. The mean and standard error (SE) of fold induction were calculated from at least three independent experiments. Similar experiments performed with the ɛ-162Luc plasmid containing the mouse GL ɛ promoter fragment as the reporter plasmid demonstrate that p50–c-Rel is able to transactivate the mouse GL ɛ promoter.

The EMSA data in Fig. 4 show that although LPS predominantly induces p50–c-Rel binding in all three B-cell lines and induces p50-p50 in 1B4.B6 cells, it also induces significant binding of p50-RelA to the CD40RR in nuclear extracts from M12.4.1 and 1B4.B6 cells. Yet LPS-induced p50-RelA does not induce GL γ1 promoter activity and transcripts. Therefore, we determined whether p50–c-Rel and p50-p50 inhibit transactivation of the promoter by p50-RelA and p50-RelB dimers by cotransfecting the −954WT reporter plasmid and the p50-RelA or p50-RelB plasmid together with the p50–c-Rel or p50-p50 plasmid. The effects of three different doses of p50–c-Rel or p50-p50 on GL γ1 promoter activity are shown in Fig. 7. It can be seen that transcription induced by p50-RelA or p50-RelB is inhibited by p50–c-Rel or p50-p50. When present at a 2:1 ratio, p50–c-Rel inhibits induction by p50-RelA or p50-RelB by 80%. It appears likely that the inhibition is due to a competition for binding to the κB sites in the GL γ1 promoter, since all of these NF-κB/Rel proteins bind the γ1 κB sites. The inability of p50–c-Rel to transactivate the GL γ1 promoter appears to be specific to this promoter, because p50–c-Rel transactivates the mouse GL ɛ promoter by about 14-fold (Fig. 7).

These data provide an explanation for the fact that the GL γ1 promoter and GL γ1 transcripts are significantly induced by CD40L but are poorly induced by LPS. LPS induces the binding of p50-RelA to the GL γ1 promoter but also induces a greater amount of p50–c-Rel and p50-p50, which do not activate the GL γ1 promoter and override the transactivation activity of the p50-RelA dimer. Although CD40L also induces binding of c-Rel and p50 to the promoter, it induces an excess of transactivating dimers. At early time points (Fig. 2A), however, the level of GL γ1 transcripts induced in 1B4.B6 cells by LPS was approximately the same as the level induced by CD40L. This appears to be consistent with the higher ratio of RelA to c-Rel at early time points after LPS addition (Fig. 3C). In contrast to LPS, CD40L induces persistent binding activity of the transactivating NF-κB/Rel dimers p50-RelA and p50-RelB.

Comparison of the mechanisms of activation of NF-κB/Rel proteins by LPS and by CD40L.

The mechanisms that mediate more persistent nuclear accumulation of RelA and RelB induced by CD40L, compared to LPS, and the mechanisms for maintenance of c-Rel activation by LPS and CD40L treatment have not been elucidated. The slow kinetics of RelB induction compared to other NF-κB/Rel proteins (Fig. 3A to C) suggests that induction of RelB may require protein synthesis. To examine this possibility, we tested the effect of CHX treatment on the induction of RelB by LPS or by CD40L. As shown in Fig. 3D, CHX treatment inhibits induction of RelB by LPS and CD40L but not of p50, RelA, or c-Rel in M12.4.1 cells. In splenic B cells, CHX treatment enhances induction of RelA and c-Rel by CD40L but it reduces the induction of RelB by CD40L (Fig. 3A). These data indicate that activation of RelB is regulated differently from activation of RelA and c-Rel and requires protein synthesis.

We next examined whether LPS and CD40L differentially target IκBα and IκBβ. The degradation of cytoplasmic IκBα and IκBβ proteins induced by LPS and CD40L in the two B-cell lines and in splenic B cells was examined by Western blot analyses. As shown in Fig. 8A to C, both LPS and CD40L cause degradation of IκBα and IκBβ proteins in these B cells, except that degradation of IκBα and IκBβ induced by CD40L in 1B4.B6 cells is not obvious, consistent with the slow kinetics of NF-κB activation induced by CD40L in this cell line. LPS or CD40L treatment in the presence of CHX leads to complete or nearly complete disappearance of IκBα and IκBβ proteins in M12.4.1 and splenic B cells (Fig. 8D), indicating that these IκB proteins are undergoing synthesis in the activated B cells.

FIG. 8.

Western blot analyses demonstrate induction of IκB degradation by CD40L and LPS. Cytoplasmic extracts (15 μg) from splenic B (A), M12.4.1 (B), and 1B4.B6 (C) cells, left untreated or treated with LPS, control supernatant (sup.), or CD40L supernatant for the indicated times, were analyzed by Western blotting. (D) Cytoplasmic extracts from M12.4.1 and splenic B cells, left untreated or treated for 6 h with control supernatant, CD40L supernatant, or LPS in the presence or absence of CHX, were used in Western blot analyses. Each Western blot was analyzed with antibodies against IκBα and IκBβ proteins sequentially.

Several cell-specific differences exist in the degradation patterns. CD40L appears to induce greater and more persistent degradation of IκBα than LPS in splenic B cells, which may explain why CD40L activates NF-κB/Rel proteins much better than LPS in splenic B cells. In 1B4.B6 cells, IκBβ is greatly reduced for at least 24 h after LPS treatment (Fig. 8C). However, in M12.4.1 cells, LPS and CD40L have similar effects on both IκBα and IκBβ (Fig. 8B), despite their differential effects on NF-κB activation. These data suggest that differential NF-κB activation by CD40L and by LPS may be partly due to targeting of different IκB proteins, but other mechanisms also exist.

DISCUSSION

Role of NF-κB/Rel proteins in IgG1 production during T-cell-dependent immune responses.

Preferential induction of the IgG1 antibody isotype during T-cell-dependent immune responses in mice appears to be mediated by CD40 signaling, as disruption of CD40L-CD40 interaction severely affects production of IgG1 but not other IgG isotypes (for a review, see reference 16). CD40 engagement contributes to the preferential IgG1 production via induction of GL γ1 transcripts by activating NF-κB/Rel proteins. NF-κB/Rel proteins have been shown to associate critically with immune responses (1). Targeted disruptions of individual NF-κB/Rel genes in mice have various effects on expression of Ig classes, providing further evidence of the important roles of NF-κB in isotype switching. Data presented in this report demonstrate that p50, RelA, and RelB are activated by CD40L, resulting in transactivation of the GL γ1 promoter. Thus, IgG1 production in response to T-dependent antigens should be reduced in mice deficient in p50, RelA, or RelB. It has been shown that mice deficient in p50 or RelB have impaired IgG1 production in response to T-dependent antigens (38, 49). The defect in IgG1 production is more severe in p50^−/− mice than in RelB^−/− mice, which can be explained by the observations that all CD40L-induced NF-κB/Rel dimers binding to the GL γ1 promoter contain p50 and that p50-RelA transactivates the promoter. However, when splenic B cells from p50^−/− mice were tested for the ability to express GL γ1 transcripts in vitro, it was found that CD40L plus IL-4 plus IL-5 induces wild-type levels of transcripts (43). It is likely that IL-4 circumvents the defect in NF-κB activation, since the GL γ1 promoter can also be activated by IL-4. In the in vivo situation, the cytokines and T-cell contact help are probably more limiting and locally delivered; therefore, each signaling pathway may be important for GL γ1 promoter activation. This may also explain why cultured splenic B cells from RelB^−/− mice produce normal levels of IgG1 in response to CD40L plus IL-4 plus IL-5 (42). However, it is still possible that p50 and RelB are involved at additional levels in class switching in vivo.

Disruption of the relA gene causes embryonic lethality (2). However, SCID mice with B cells reconstituted by transplantation of RelA^−/− fetal liver cells show a 10-fold decrease in serum IgG1, but not in all isotypes, compared to SCID mice receiving normal fetal liver cells. IgG1 production in response to T-dependent antigens was not investigated (10).

In c-Rel-deficient mice, cytokine production and T-cell functions are affected, so the effects on antibody production are probably indirect (22). It has recently been reported that B cells from mice with a mutated c-rel gene that contains the DNA-binding domain but no transactivation domain do not express GL γ1 transcripts in response to LPS and IL-4 (52). These results may seem to contradict our data, but it is likely that the c-Rel DNA-binding domain competes with the transactivating NF-κB/Rel dimers for binding to the GL γ1 promoter and thereby inhibits promoter activity.

CD40L and LPS have different patterns of NF-κB activation.

Induction of nuclear translocation of NF-κB/Rel proteins by a variety of NF-κB inducers is regulated by degradation of IκB proteins and can be transient or persistent (1). LPS has been shown to induce NF-κB binding activity more persistently than phorbol myristate acetate and TNF-α (46). However, our data demonstrate that LPS persistently activates p50 and c-Rel, but not RelA or RelB, in B cells, indicating that persistent induction can be restricted to certain NF-κB/Rel proteins and that nuclear accumulation of individual NF-κB/Rel proteins is regulated differentially. By contrast, CD40L activates all NF-κB/Rel proteins in splenic B cells and in the two B-cell lines we examined for more than 24 h (Fig. 3 and data not shown). Neumann et al. (30) reported similar results, showing that CD40L expressed on L cells induces RelA for 24 h and p50, c-Rel, and RelB for up to 48 h in splenic B cells.

The mechanisms that mediate more persistent induction of RelA and RelB by CD40L and the mechanisms for persistent induction of c-Rel by LPS and CD40L treatment need to be further investigated. Analyses of the degradation of IκB proteins induced by CD40L and LPS indicate that targeting of different IκB proteins may contribute to differential NF-κB activation. However, CD40L and LPS induce similar degradation kinetics of IκBα and IκBβ in M12.4.1 cells, indicating that other mechanisms must regulate differential NF-κB activation by these inducers. Other IκB proteins that we have not examined may be differentially regulated (1, 50). Another possibility is that nuclear RelA and RelB induced by CD40L, as compared to LPS, may be more stable.

The mechanisms that mediate induction of RelB appear to differ from those activating RelA and RelB and remain to be elucidated. Although CHX treatment inhibits nuclear translocation of RelB, we found that the levels of RelB mRNA and cytoplasmic RelB protein are not changed after 12 h of CD40L treatment in M12.4.1 cells; thus, it is unlikely that synthesis of RelB protein mediates RelB induction at this time (data not shown). It has been shown, however, that after 24 h of CD40L stimulation of splenic B cells, RelB mRNA levels are induced (30). One possible mechanism for the unique characteristics of RelB induction is that a newly synthesized RelB transporter(s) and/or signaling protein is required for nuclear translocation and activation of RelB.

Positive and negative regulatory effects of c-Rel on gene expression.

Gene expression can be positively or negatively regulated by c-Rel, depending on the gene and cell type. In c-Rel-deficient mice, production of IL-3 and GM-CSF by T cells and production of TNF-α and inducible nitric oxide synthase by resident peritoneal macrophages are significantly reduced, indicating that c-Rel can transactivate gene expression (12, 17). In contrast, the expression of GM-CSF, G-CSF, and IL-6 in resident peritoneal macrophages is increased in c-Rel-deficient mice (17), indicating that c-Rel also plays a negative role in controlling gene expression. Further evidence for a negative role is the finding that expression of c-Rel inhibits RelA-mediated transactivation of the long terminal repeat of human immunodeficiency virus (9).

By overexpression of NF-κB/Rel fusion proteins, we found that the p50–c-Rel dimer has different effects on the mouse GL γ1 and ɛ promoters. The p50–c-Rel dimer poorly induces the GL γ1 promoter and suppresses transactivation of the GL γ1 promoter by p50-RelA and p50-RelB dimers. Since the p50–c-Rel heterodimer and p50 homodimer bind well to the CD40RR of the GL γ1 promoter, repression appears to be mediated by competition with transactivating NF-κB/Rel dimers for DNA binding. In contrast, GL ɛ promoter activity is efficiently induced by p50–c-Rel fusion protein, albeit still not as well as by p50-RelA.

The transactivation activity of c-Rel may be mediated by interaction with other transcription factors. For example, Wang et al. (47) reported that coexpression of c-Rel together with NF-ATc protein, but not c-Rel or NF-ATc alone, strongly induces IL-2 promoter activity. RelA did not appear to substitute for c-Rel. Therefore, c-Rel may preferentially activate promoters that also bind NF-ATc.

The preferential induction of c-Rel might cause LPS to inhibit induction of the GL γ1 promoter by CD40L. However, we found that when added together with CD40L, LPS does not significantly reduce induction of promoter activity by CD40L in M12.4.1 cells (25). p50-RelA and p50-RelB were found to be the dominant NF-κB/Rel dimers binding to the GL γ1 promoter in nuclear extracts from M12.4.1 cells treated with LPS and CD40L (data not shown). Surprisingly, however, the combination of LPS and CD40L synergistically induces the endogenous GL γ1 transcripts in 1B4.B6 cells, although p50–c-Rel is still the predominant NF-κB/Rel dimer induced by LPS plus CD40L in these cells (data not shown). The synergistic effect of LPS and CD40L might be explained by activation of transcription via sequences outside the promoter segment present in the luciferase reporter gene we use, perhaps by interaction of the p50–c-Rel dimer with another transcription factor(s) induced by LPS plus CD40L.

In conclusion, the ability of CD40 signaling to induce GL γ1 transcription contributes to the preferential production of IgG1 production during T-dependent immune responses. Persistent activation of transactivating NF-κB/Rel dimers in excess over nontransactivating NF-κB/Rel dimers mediates induction of GL γ1 transcripts by CD40L. In contrast, LPS predominantly activates nontransactivating NF-κB/Rel dimers, explaining why LPS in the absence of T-cell help does not induce IgG1 production in B cells.