Differential Regulation of Mouse Germline Ig γ1 and ε Promoters by IL-4 and CD40¹

Upon activation by Ag or mitogens, IgM⁺ B cells undergo Ig class switching, resulting in expression of a different heavy chain constant region (C_H) gene, or isotype while maintaining expression of the same variable region gene (reviewed in Refs. 1, 2, 3). Because the C_H region determines the Ab effector function, class switching allows the humoral immune response to adaptively respond to a variety of different infectious organisms. Class switching is mediated by DNA recombination occurring between switch (S) region sequences residing 5′ to each C_H gene, except Cδ (4, 5).

Numerous studies have established that transcription of unrearranged C_H genes occurs before switch recombination, producing what are termed germline (GL)³ or switch transcripts (reviewed in Ref. 2). Transcription initiates at an I exon located 5′ to each S region and continues through the S region and C_H gene. Splicing removes the S region sequences and joins the I exon with the C_H exons. Deletions of various portions of the I exons and their upstream regulatory elements by gene targeting experiments have demonstrated that GL transcripts are required for switch recombination (6, 7, 8, 9, 10). However, the role of GL transcripts in directing class switch recombination is unknown.

In this manuscript, we examine mechanisms regulating differential expression of the GL transcripts required for IgG1 and IgE class switching in the mouse. IgG1 can bind and activate complement, binds the low-affinity FcγRIII receptor on macrophages, neutrophils, mast cells and NK cells, and is protective against bacterial and viral infections (3). IgE binds to high-affinity receptors (FcεRI) on mast cells and basophils and induces degranulation and cytokine production by these cells when engaged with Ag, helping to eliminate parasitic helminths (11). However, IgE causes allergic responses, and in industrial societies is often more dangerous than protective.

IgG1 and IgE are produced in response to T-dependent Ags, although IgG1 in much greater abundance than IgE. Both isotypes are produced during Th2 immune responses, as they are induced by the cytokine IL-4 (12, 13). Furthermore, T cell contact help, primarily mediated by CD40-CD40 ligand (L) interaction, is important for B cell activation, proliferation, and isotype switching during T-dependent immune responses (14, 15, 16). Ab to CD40 or CD40L synergizes with IL-4 to induce switching to IgG1 and IgE in cultured mouse splenic B cells (17, 18) and expression of certain IgG subclasses and IgE in cultured human B cells (19, 20). Although IL-4 induces both IgG1 and IgE, these isotypes differ in their dependence on IL-4. Switching to IgG1 is only partially reduced in T cell-dependent immune responses in IL-4-deficient mice, whereas IgE switching is undetectable or only weakly induced by long-term infection with parasites (21, 22, 23, 24).

IL-4 directs class switch recombination to IgG1 and IgE by inducing GL γ1 and ε transcripts in activated B cells (25, 26, 27). CD40L or Ab to CD40 synergizes with IL-4 to further induce GL ε and mouse γ1 transcripts in cultured mouse splenic B cells (28, 29, 30, 31, 32). In the absence of IL-4, CD40 signaling induces modest levels of GL γ1 and ε transcripts, but γ1 transcripts appear to be more inducible than ε transcripts (18, 31, 32).

In addition to evidence suggesting that the GL γ1 transcripts are more inducible by CD40 signaling than are ε transcripts, cross-linking of surface Ig in vitro has been shown to induce the promoter for GL γ1 transcripts (33) and anti-IgD conjugated with dextran (anti-δ dextran) has been shown to induce endogenous γ1 transcripts (34). By contrast, induction of GLε transcripts by IL-4 and LPS is inhibited by treatment of splenic B cells with anti-δ dextran (34).

Transient reporter gene assays have been used to examine the mechanism of transcriptional regulation of GL γ1 and ε RNA by CD40L and IL-4. Similar to the endogenous transcripts, treatment with CD40L and IL-4 activates transcription from both promoters, and CD40L alone activates the mouse GL γ1 promoter somewhat better than the ε promoter (35, 36, 37). In addition, the ε promoter can be induced by IL-4 alone, whereas the γ1 promoter cannot be (33, 36, 38). Therefore, transcription of GL ε RNA appears to be more inducible by IL-4, but also more dependent on IL-4, than transcription of γ1 RNA.

In this report, we compare the promoters for GL γ1 and ε transcripts to understand the differential regulation of IgG1 and IgE expression. We address the question of why the GL ε promoter is more responsive to IL-4 than is the γ1 promoter. The IL-4-responsive regions of the GL γ1 and ε promoters are similar, each containing a binding site for the IL-4-inducible transcription factor Stat6, located immediately adjacent to a binding site for a basic region leucine zipper (bZip) family protein that is important for IL-4 induction. However, the placement of the two sites relative to each other and the sequences of the two elements differ (see Fig. 1, A and B). We ask whether the placement of the two elements or the different sequences of the elements are more important for differential regulation of the GL γ1 and ε transcription. Our data also address the question of why the ε promoter is more dependent on IL-4 than is the γ1 promoter.

Splenic B cells were isolated from 8- to 10-wk-old BALB/c or 129 × C57BL/6 mice by depletion of T cells as described previously (39). Mouse splenic B cells and the B cell lines M12.4.1 and A20.3 were maintained at 37°C in 5% CO₂ incubator in RPMI 1640 medium (BioWhittaker, Walkersville, MD) supplemented with 10% FCS (HyClone, Logan, UT or BioWhittaker), 0.1 mM nonessential amino acids, 100 U/ml penicillin, 100 μg/ml streptomycin, 100 μg/ml kanamycin, 1 mM sodium pyruvate, 2 mM l-glutamine (all from Life Technologies, Grand Island, NY), and 50 μM 2-ME (Sigma, St. Louis, MO).

Recombinant mouse IL-4 (a gift from W. E. Paul, National Institutes of Health, Bethesda, MD) was a culture supernatant prepared from insect cells infected with recombinant baculoviruses (40). The source of CD40L was a supernatant containing CD40L-CD8α fusion protein secreted from stably transfected J558L myeloma cells cultured in complement RPMI 1640 medium-10% FCS (15, 35). Supernatant from nontransfected J558L cell cultures, used at the same concentration, was the control supernatant. The amounts used are indicated in the figure legends. Anti-δ-dextran (3 ng/ml; gift from C. Snapper, Uniformed Services University of the Health Sciences, Bethesda, MD) or LPS (50 μg/ml, 055:B5; Sigma) dissolved in RPMI 1640 medium was added to splenic B cell cultures at the initiation of the culture.

The following oligonucleotide sequences (upper strand) were used for DNA probes or competitor fragments in EMSAs. Lower case letters indicate mutated nucleotides. Binding sites for transcription factors C/EBP, CREB/activating transcription factor (ATF), or AP-1 are underlined. Stat6 binding sites are in bold face: γ1(S + B), 5′-ACACATTCACATGAAGTAATCTAAG-3′; γ1Stat6, 5′-ACACATTCACATGAAGagtTCTAAG-3′; γ 1bZip, 5′-ACACAggttCATGAAGTAATCTAAG-3′; ε(S+B), 5′-TGCCTTAGTCAACTTCCCAAGAACAGA-3′; AP-1 consensus (CS), 5′-AGCTTGGTGACTCATCCG-3′; C/EBP CS, 5′-TGCAGATTGCGCAATCTGCA-3′; CREB/ATF CS, 5′-AGAGATTGCCTGACGTCAGAGACT-3′; Stat6, 5′-ACTTCCCAAGAACA-3′; Bcl6 CS, 5′-GAAAATTCCTAGAAAGCATA-3′; octamer-binding protein (Oct), 5′-TGTCGAATGCAAATCACTAGAA-3′.

The luciferase reporter plasmid γ1Luc (=Luc4) containing the mouse GL γ1-148/+202 promoter segment and the luciferase reporter plasmid (εLuc) containing the mouse GL ε promoter −162/+53 segment have been described previously (33, 38). For construction of plasmid γ1Non-ov, γ1Inv, and γ1Luc(ε), DNA fragments were amplified by PCR with γ1Luc as template with forward primer: NonF, 5′-CAGGGTACCGCCTCACCCTCACCCACACATTCACATGAATGAAGTAATCTAAGTCAGGTTTG-3′; InvF, 5′-CAGGGTACCGCCTCACCCTCACCCACACATGAAGTAATTTCACATGAACTAAGTCAGGTTTG-3′; g1eF, 5′-CAGGGTACCGCCTCACCCTCACCCACACATTAGTCAACTTCCCAAGAACTAAGTCAGGTTT-3′. The KpnI sites used for cloning are underlined. One reverse primer was used: g1R, 5′-CAAGCTgAGATCTGGAAG-3′ (BglII site is underlined). The amplified fragments were digested with KpnI and BglII and cloned into the pXP2 luciferase vector (41) digested with the same enzymes. To construct the reporter plasmid γ1Luc(εB) and γ1Luc(εS), DNA fragments were amplified from plasmid γ1Inv with g1R as the reverse primer paired with one of the following forward primer: γ1Luc(εB)F, 5′-CAGGGTACCGCCTCACCCTCACCCACACAttagtcaaTTCACATGAACTAAGTC-3′ or γ1Luc(eS)F, 5′-CAGGGTACCGCCTCACCCTCACCCACACATGAAGTAATttcccaagaaCTAAGTC3′. The lowercase letters represent the εAP-1 site sequence (in g1eBF) or εSTAT6 site (in g1eSF). The amplified fragments were digested with KpnI and BglII and inserted into pXP2. Similarly, to construct the reporter plasmid εLuc(γ1), a DNA fragment was amplified from εLuc template with a primer pair as follows: eg1F, 5′-GGCAGGCCTCACCTGAGACCcCACTGTGCCTTCACATGAAGTAATCAGAATCAAAAGGGAAC-3′ (StuI site sequence is underlined) and eR, 5′-CCAAGCTgAGATCTGTGC (BglII site sequence is underlined). The amplified fragments were digested with StuI-BglII and inserted into εLuc, digested with the same enzymes.

Other mutated reporter plasmids were generated with the QuickChange Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer’s protocol. Three basepair replacement mutations in the bZip site of the γ1 promoter were obtained using the mutagenic oligonucleotide mbZipF: 5′-ACACATTCACATGAAGagtTCTAAGTCAGGTTTG-3′. Only the forward primer sequence is shown. The γ1 promoter containing three mutated NF-κB sites, m3κB, was created by sequentially replacing each of the three NF-κB sites with the following mutagenic oligonucleotides: mκB-1F, 5′-CTAAGTCAGGTTTtGACTCgagCTCACCCTCTGAC-3′; mκB-2F, 5′- CCTCTGACACAGAActgCagAGAATGAAGGGGAAC-3′; mκB-3F, 5′-CCCCAGAATGAAttctAACCCTGTCAGGAAATG-3′. The underlined sequences represent individual NF-κB sites, and the lowercase letters represent the mutagenic nucleotides. The reporter plasmid γ1Luc-mbZip-m3κB, containing mutations in both the bZip site and 3-κB sites was similarly generated with the primer mbZip on the m3κB γ1 reporter plasmid. DNA sequences of the wild-type and mutant promoters in the reporter plasmids were verified by sequencing the entire γ1 or ε sequence and flanking vector sequences. Luciferase reporter plasmids containing a minimal alkaline phosphatase promoter linked upstream with (E4APLuc) or without (APLuc) four copies of a C/EBP binding site (42) were provided by C. Cooper (University of Massachusetts Medical School, Worcester MA).

The eukaryotic expression plasmids, cloned in pcDNA3, expressing CEBPβ, C/EBPγ, c-Fos, JunB, or c-Jun were described previously (32). Expression plasmids pcDNA3-ATF2 and the dominant-negative pcDNA3-ATF2dn were provided by Phyllis LuValle (University of Calgary, AB, Canada) (43). The expression plasmid for mouse ATF4 was generated from cDNA obtained from I.29μ B cell RNA by RT-PCR with ATF4 forward primer 5′-ATGGGATCCACAACCATGACCGAGATGAG-3′, and reverse primer 5′-GAGGAATTCACAAAGCACCTGACTAC-3′. The underlined letters represent BamHI (in forward primer) or EcoRI (in reverse primer) sites. The amplified DNA fragment was digested with BamHI and EcoRI and inserted into pcDNA3, cut by the same enzymes. DNA sequencing confirmed that the cloned gene has exactly the same sequence as the mouse ATF4 gene in GenBank (accession number M94087) (44). The bZip domain of C/EBPβ (amino acids 187–296) was amplified by PCR from the mouse C/EBPβ expression plasmid with the upstream primer 5′-AGGAATTCCACCATGGCGCCCGGCACGCCGAG-3′ and downstream primer 5′-CGCGCCGCGCTAGCAGTG-3′. The underlined nucleotides indicate the EcoRI and NheI restriction sites. The resulting product was digested by EcoRI and NheI and ligated into pcDNA3, cut by EcoRI and XbaI.

All of the plasmids used for transient transfection were prepared using Qiagen kits (Valencia, CA). Mouse M12.4.1 or A20.3 cells mixed with 30 μg of reporter plasmid and the indicated amount of expression plasmids, along with 1 μg of β-galactosidase (β-gal) reporter plasmid pPGK-gal (Ref. 45 ; except in the experiment shown in Fig. 1) in 1 ml of RPMI 1640 were electroporated at 1250 μF/300 V and split into equal fractions. Aliquoted cells were left untreated or treated for 12 h with IL-4, CD40L, or the phorbol ester PMA, as indicated, in complete RPMI 1640 medium. Cell lysates and luciferase assays were prepared as described previously (36). The β-gal assays were performed as described on the samples that were not treated with inducers (46). Luciferase values were normalized relative to the internal β-gal activity.

Nuclear extracts of splenic B cells and of M12.4.1 B cells were prepared as described previously (36). DNA binding reactions were performed at room temperature for 15–20 min in 15-μl reaction volumes containing nuclear extracts from splenic B cells or from M12.4.1 B cells (3 μg), 4 nmol of ³²P-end-labeled DNA probe, 2 μg of poly(dI-dC), 10 mM Tris-HCl (pH 7.6), 1 mM MgCl₂, 100 mM NaCl, 1.2% (v/v) glycerol, 5 mM KCl, 0.1 mM EDTA, and 1 mM DTT. For DNA competition experiments, a 100-fold molar excess (unless otherwise indicated) of unlabeled competitor oligonucleotides relative to the labeled probe was incubated in the binding mixture for 20 min before addition of the ³²P-labeled probe. For Ab supershift experiments, 1 μl of the indicated specific Ab was added to the binding reaction. The binding reaction mixtures were electrophoresed in 5% native polyacrylamide gels with 0.5× Tris-borate-EDTA buffer, followed by autoradiography. Abs against mouse CREB-1 (sc-186X, cross reactive with ATF1 and CREM1, as indicated by the manufacturer), ATF4 (CREB-2; sc-200X), ATF2 (sc-187X), ATF3 (sc-188X), C/EBPα (sc-061X), C/EBPβ (sc-150X), C/EBPγ (sc-7659X), and STAT6 (sc-981X) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).

Thirty micrograms of nuclear protein from splenic B cells, M12.4.1 or BclI-3B3 was fractionated on 15% reducing SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membranes. Membranes were first incubated with the Ab against ATF2 or C/EBPβ in TBST buffer (20 mM Tris-HCl (pH 7.5), 137 mM NaCl, and 0.05% Tween 20) in the presence of 5% nonfat milk powder. Upon washing with TBST, blots were incubated with HRP-conjugated second Ab. The Abs and HRP conjugate were purchased from Santa Cruz Biotechnology. The immunoreactive bands were revealed on film by using the SuperSignal chemiluminescent substrate (Pierce, Rockford, IL) according to the manufacturer’s instructions.

The IL-4-responsive region of the mouse GL ε promoter has been studied by a combination of reporter gene assays and EMSAs. These experiments have demonstrated that two adjacent sequence elements are required for IL-4 induction of the mouse GL ε promoter, a binding site for Stat6 and a site for AP-1 (Fos and Jun) transcription factors (consensus site (CS), TGAGTCA) (32, 36, 38 , and 47). Fig. 1 B presents the sequence of the IL-4-responsive region of the GL ε promoter. If either the Stat6 or AP-1 element is mutated, IL-4 induction is completely eliminated (32, 36, 38). Furthermore, cotransfection of various AP-1 proteins along with the ε reporter plasmid into the I.29μ B cell line greatly activates the promoter in IL-4-treated cells (32). Although the ε AP-1 site can bind bacterially produced recombinant C/EBPβ(NF-IL-6) and C/EBPγ(Ig/EBP), this site binds AP-1 but not C/EBP when EMSAs are performed with B cell nuclear extracts. Furthermore, C/EBP proteins do not activate transcription of the mouse GL ε promoter (32). Both AP-1 and C/EBP are bZip transcription factors, a family that also includes ATF/CREB proteins. The three different families of bZip proteins, AP-1, C/EBP, and ATF, have related but not identical binding sites, although they often are found to bind each other’s sites with lower affinity (48, 49)

The elements of the GL γ1 promoter that are required for IL-4 responsiveness have not been defined, although a Stat6 binding element has been identified within the IL-4-responsive region (33, 50, 51, 52). Overlapping the Stat6 site is an element, TGAAGTAAT, that binds C/EBPβ and C/EBPγ in B cell nuclear extracts, as assessed by Ab supershift assays (52). Mutations created within the C/EBP site suggested its importance for promoter expression, although because of the large numbers of nucleotides mutated, the data were not definitive (33). Furthermore, cotransfection experiments to assess whether C/EBP proteins activate the GL γ1 promoter have not been reported. Thus, it is unknown whether this element is required for promoter expression and for IL-4 induction, nor is it known which proteins might activate transcription at this site. Fig. 1 A shows the sequence of the GL γ1 promoter region containing the Stat6 and C/EBP elements. To indicate the possibility that other proteins may function via the latter element, we label this element bZip.

To determine whether the differences in the Stat6 and bZip elements of these promoters affect induction of the GL γ1 and ε transcripts by IL-4 and CD40 signaling, we systematically interchanged the consensus elements of this region between the two promoters. The activity of each of the reporter plasmids was tested by transient reporter gene assays in the B cell line M12.4.1. This cell line was chosen because it supports CD40L-inducible expression of both of these promoters, and IL-4-inducible expression of the GL ε promoter. However, the level of IL-4 induction of the GL γ1 promoter in this cell line is very low. The IL-4-responsive region of the γ1 promoter was defined previously in two B cell lines, L10A6.2 and A20.3, in which IL-4 synergizes with phorbol ester to induce the promoter (Fig. 2,B and Ref. 33). Phorbol ester delivers a signal that partially mimics the signals produced by cross-linking surface Ig. Ab against surface Ig also synergizes with IL-4 to activate the GL γ1 promoter (33). By contrast, in M12.4.1 cells, the γ1 promoter is poorly inducible by phorbol ester or by the combination of phorbol ester and IL-4 (Fig. 2 A). We could not use A20.3 because the GL ε promoter is not expressed well in these cells, nor could we use L10A6.2 because the γ1 promoter is not inducible by CD40L in these cells (data not shown). Presumably, the GL ε and γ1 promoters respond differently among cell lines because of differences in receptors, signaling molecules, or transcription factors, and different responses between cell lines are not unique to these promoters. Because of this, it is important to use the same cell line for comparative studies of the two promoters.

The luciferase reporter plasmid driven by the wild-type GL γ1 promoter, containing nucleotides −148/+202, relative to the first RNA initiation site, was transfected into M12.4.1 along with an internal control plasmid, PGK-βgal. Cells were divided into six aliquots and placed in medium without additions or treated with IL-4, soluble CD40L, or with the combinations shown in Fig. 1. After 12 h, luciferase and β-gal activities were measured. It can be seen in Fig. 1 A (right) that the γ1 promoter can be induced by CD40L by 5-fold, but not by IL-4 alone, although their combination induces slightly more than CD40L alone.

The wild-type GL ε promoter −162/+53 segment, assayed identically, is more inducible by IL-4, 5-fold by IL-4 alone, but less inducible by CD40L alone (2-fold; Fig. 1,B and Ref. 36). The combination is synergistic, giving a 32-fold increase relative to the medium control. However, even in the presence of IL-4 + CD40L, the overall luciferase activity obtained from the GL ε promoter is less than that of the γ1 promoter (Fig. 1 B, middle).

We hypothesized that the differences in the Stat6 and bZip elements of these two promoters might affect the different levels of activity and the differential responsiveness of the γ1 and ε promoters to CD40L and IL-4. There are three obvious differences between these two promoters in this region. The Stat6 and bZip elements differ in their positions relative to each other and are overlapping in the γ1 promoter and separate in the ε promoter. In addition, the sequences of both the Stat6 and bZip elements of the two promoters differ.

To make the subsequent experiments easier, we first tested whether separating the γ1 bZip and Stat6 elements so that they would not overlap would affect promoter activity (Fig. 1,C). This change has a minimal effect on the promoter activity, although it reduces the responsiveness to the combination of IL-4 and CD40L. We next tested whether responsiveness of the γ1 promoter would be altered if the positions of the Stat6 and bZip elements were exchanged relative to each so that now the Stat6 site is located 3′ to the bZip site. As shown in Fig. 1 D, this had only a small effect, indicating that the difference between the γ1 and ε promoter is not attributable to the different relative positions of the Stat6 and bZip elements.

These data suggested that the specific sequences of the Stat6 and/or bZip elements are likely to be important in the differential responsiveness. We tested this hypothesis by replacing the ε elements with the sequences of the γ1 elements and vice-versa. Fig. 1 E shows that replacing the Stat6 and AP-1 sites of the ε promoter with the two γ1 elements results in a promoter with very little activity. By contrast, substituting the two ε elements into the γ1 promoter results in a promoter that is highly inducible by IL-4 alone (18-fold) and 4-fold by CD40L alone, with the combination giving a total induction of 47-fold. Also, the overall activity of the chimeric promoter is 10 times greater than either wild-type promoter. Altogether, these data indicate that the IL-4-responsive region of the ε promoter is responsible for the greater inducibility of the ε promoter and that this is attributable to the particular sequences of this region. Furthermore, the data suggest that the remainder of the γ1 promoter has additional elements that provide a greater overall level of activity than that of the GL ε promoter. This result is consistent with data indicating that the ε promoter is more dependent on IL-4 than the γ1 promoter, whereas the γ1 promoter can also be activated by additional transcription factors that do not depend on IL-4 signaling.

We next asked whether the ε Stat6 and ε AP-1 binding sites are both required for this increased responsiveness by replacing only the γ1 bZip site (Fig. 1,G) or only the γ1 Stat6 site (Fig. 1 H). It can be seen that both ε elements contribute to the increased responsiveness, as neither of these single-element replacements is as active as the promoter containing both ε elements, although each is slightly more inducible by IL-4 than the wild-type γ1 promoter.

One possible explanation for the greater IL-4 inducibility of the GL ε promoter relative to the γ1 promoter is that the ε Stat6 element has a higher affinity for Stat6 than the γ1 element. Although both Stat6 elements match the consensus binding site, TTCN₄GAA, they differ in the central and flanking nucleotides. The relative affinity of the ε and γ1 Stat6 sites was examined by performing competition EMSAs with a double-stranded oligonucleotide containing the ε(Stat6 + AP-1) segment as the probe (Fig. 3,A). Competition was performed with the identical oligonucleotide, or with other oligonucleotides containing the γ1(Stat6 + bZip) elements, or chimeric oligonucleotides containing the ε Stat6 + γ1 bZip sites and vice versa (see Fig. 3,A). Fig. 3,B demonstrates that the ε(Stat6 + AP-1) probe binds an IL-4 inducible factor in M12.4.1 nuclear extracts, shown by Ab supershift assays to be Stat6 (36), and that this binding is competed by the unlabeled probe (lanes 2–7). The γ1(Stat6 + bZip) oligonucleotide competes poorly (lanes 9–12). Use of chimeric oligonucleotides demonstrates that the Stat6 element is responsible for the differing abilities to compete and that spacer nucleotides account for the difference (lanes 13–17 and 18–22). Quantitation of the results by densitometry demonstrates that by this assay the ε Stat6 element has an ∼10-fold greater affinity for Stat6 than does the γ1 element (Fig. 3 C). The dramatic influence of the sequence of the spacer nucleotides on binding affinity is consistent with the recent report of Ehret et al. (53). In fact, according to this study, the GL ε Stat6 site is nearly an optimal binding site for Stat6. The higher affinity for Stat6 is consistent with the greater responsiveness of the GL ε promoter than the γ1 promoter to IL-4.

To determine whether the bZip element is important for expression of the γ1 promoter, a mutation was created in the bZip element which does not disrupt the Stat6 binding site, as shown by competition EMSA (data not shown). Fig. 2,C shows the sequence of the mutated promoter (γ1Luc-mbZip). This reporter plasmid was tested in comparison to the wild-type γ1 promoter by transient transfection into two B cell lines (Fig. 2). The bZip mutation reduces activity of the promoter by 2.2-fold in M12.4.1 and by 5-fold in A20.3 cells. Although the overall level of promoter activity is reduced, the bZip element does not appear to be essential for induction by CD40L, CD40L + IL-4, or by phorbol ester. Thus, the bZip site of the γ1 promoter does not play the same role as the AP-1 site of the GL ε promoter, in which it has been shown to be essential for both IL-4 inducibility and overall promoter activity in both M12.4.1 and I.29μ B cells (32, 38).

It is clear from data shown in both Figs. 1 and 2 that the role of the bZip element of the γ1 promoter differs from the AP-1 site of the ε promoter. To explain this difference, it is important to identify the proteins that bind this site in B cell nuclear extracts. It has previously been shown that the γ1 bZip site binds C/EBPβ and C/EBPγ (52). To confirm this result and to determine whether additional proteins may also bind, we performed a series of competition EMSAs using the γ1(Stat6 + bZip) probe shown in Fig. 3,A. The EMSA results shown in Fig. 4,A demonstrate that at least six specific complexes form with this probe using nuclear extracts from unstimulated M12.4.1 cells. Stimulation of cells with LPS or CD40L does not change the complexes formed (Fig. 5,A). Addition of 100-fold excess of unlabeled probe (Fig. 4 A, lane 5) or an oligonucleotide containing the γ1 bZip site (and a mutated Stat6 site, lane 3) greatly inhibits five of the six complexes. The γ1 Stat6 (+mutated bZip element) oligo does not compete (lane 4), nor does the ε(Stat6 + AP-1) oligo (lane 6). Competition with an oligo having the consensus binding site for ATF/CREB (TGACGTCA) eliminates five of the complexes (lane 8), and competition with an oligo having the consensus C/EBP site (TTGCGCAA) competes well with complexes 2, 4 and 6 (lane 9). Oligos containing consensus elements for AP-1 (TGACTCA), Stat6, and Bcl6 do not compete (lanes 7, 10, and 11).

Similar competition experiments were performed with nuclear extracts from splenic B cells that had been treated with CD40L for 24 h (Fig. 4 B). The results are similar to those obtained with M12.4.1 extracts. Together these data suggest that the γ1 bZip site binds both ATF/CREB and C/EBP proteins in M12.4.1 and CD40L-activated splenic B cells.

Ab supershift EMSAs were performed to confirm the binding of ATF/CREB and C/EBP proteins and to identify the specific proteins that bind. Ab to ATF2 supershifts the major complex 1 formed with extracts from both M12.4.1- and CD40L-treated splenic B cells, although in splenic B cell extracts, an additional complex 1b is not affected (Fig. 4, C and D, lane 4). In addition, ATF2 Ab appears to partially inhibit complex 2. ATF-4 Ab forms a weak supershifted complex with M12.4.1 extracts but not with splenic B cell extracts (lane 3), but no other ATF/CREB Ab has an effect. Ab for C/EBPβ supershifts complex 6 formed with M12.4.1 extracts and probably also with splenic extracts (lane 7). Ab to C/EBPγ inhibits complexes 2 and 6 formed with both extracts (lane 8). These data are consistent with previous published data indicating that C/EBPβ and C/EBPγ form multiple complexes with the GL γ1 promoter (52), but in addition indicate that ATF2 forms the major complex 1 that binds the γ1 bZip site. Complex 1 could consist of ATF2 homodimers or a heterodimer with another protein, although no other Ab we tested affects this complex. Although we failed to identify the proteins in complex 1b detected with splenic B cell extracts, it may contain a member of the ATF/CREB family, because it is competed by the oligo containing the consensus ATF/CREB element (Fig. 4 B, lane 8).

The finding that ATF2 binding activity is not inducible by LPS or CD40L in nuclear extracts from M12.4.1 or A20.3 cells (Fig. 5,A and data not shown) is consistent with the finding that mutation of the bZip element in the reporter assays reduces overall expression but not inducibility in these cells (Fig. 2). However, because both of these cell lines are comparable to activated B cells, unlike freshly isolated splenic B cells, which mostly consist of resting B cells, we wished to determine whether ATF2 binding activity would be induced in splenic B cells by reagents that stimulate IgG1 switching. Splenic B cells were treated for 24 h with LPS, IL-4, CD40L, or anti-IgD conjugated with dextran (anti-δ dextran). As shown in Fig. 5 B, treatment with each of these reagents, except IL-4, induces ATF2 binding (complex 1) and also complex 1b. However, C/EBPβ or C/EBPγ binding activities are not induced. This experiment was performed three times with similar results. As expected, IL-4 induces binding of Stat6. The binding activity of Oct 1 in these extracts is shown below as a control for extract quality and loading. The finding that ATF2 binding activity is induced by 24 h of treatment of splenic B cells with LPS or with anti-δ dextran is consistent with results reported by Feuerstein et al. (54).

To attempt to determine whether the increase in binding activity of ATF2 is attributable to an increase in protein levels, we examined the levels of ATF2 in splenic B cell nuclear extracts by Western blotting (Fig. 5,C). Although freshly isolated splenic B cells do not have ATF2 binding activity, they do show constitutive levels of protein. Furthermore, the 3- to 6-fold induction of ATF2 protein levels observed by Western blotting in cells treated with LPS, anti-δ dextran, or CD40L is not nearly as great as the induction of binding activity shown by EMSA. These data suggest that increased binding activity is not primarily attributable to an increase in ATF2 levels. By contrast, C/EBPβ protein levels are nearly invariant (Fig. 5 C).

To determine whether ATF2 activates the GL γ1 promoter when it binds the bZip site, we tested the effect of cotransfection of an expression plasmid for ATF2 along with GL γ1 reporter plasmids into M12.4.1 and A20.3 cells. Two γ1 reporter plasmids, containing either the wild-type sequence or the mutated bZip element, were tested. Overexpression of ATF2 increases the activity of the GL γ1 promoter by about 2-fold in both M12.4.1 and A20.3 cells, and this induction depends on the bZip element (Fig. 6, A and B). We also tested the effect of overexpression of a dominant-negative form of ATF2, which lacks the transactivation domain (43), and found it inhibits the wild-type promoter but has no effect on the mbZip promoter. These data demonstrate that ATF2 activates the GL γ1 promoter and that this most likely occurs via the binding of ATF2 to the bZip site.

Because a small amount of a complex containing ATF4 also is detected in the EMSA supershift experiments performed with M12.4.1 nuclear extracts, we tested overexpression of an expression plasmid for ATF4 and found it has little or no effect on γ1 promoter expression (Fig. 6 A). The combination of equal amounts of ATF2 and ATF4 expression plasmid yields results similar to ATF2 alone (data not shown). These data and the lack of detection of an ATF4 complex with splenic B cell extracts suggest that ATF4 probably is not involved in regulating transcription of the GL γ1 promoter.

Finally, we tested the effect of overexpression of C/EBPβ and C/EBPγ on activity of the wild-type and mbZip reporter plasmids. As shown in Fig. 6,C, both C/EBPβ and C/EBPγ inhibit reporter activity from plasmids containing either the wild-type or mbZip promoters in M12.4.1. These results indicate that both C/EBP proteins inhibit rather than stimulate activity of the GL γ1 promoter, and that this inhibition does not require the bZip element. However, C/EBPβ and C/EBPγ do not behave identically, as C/EBPβ inhibits induction by CD40L in addition to reducing the overall level of activity, whereas C/EBPγ does not eliminate induction. These results are surprising because although C/EBPγ previously has been shown to inhibit transcription, C/EBPβ is generally found to be a transcriptional activator (55). Furthermore, although these proteins bind the bZip site in the γ1 promoter, data in Fig. 6,C indicate they do not require this site for their inhibitory activity. Note that the C/EBPβ expression plasmid we used is able to activate transcription on a reporter plasmid regulated by four tandem C/EBP binding sites (Fig. 6 D).

The ATF2 binding site is important for expression of the GL γ1 promoter, but it does not contribute to inducibility of the promoter by CD40L, anti-IgM, PMA, or the combination of CD40L and IL-4 in M12.4.1 and A20.3 cells (Fig. 2). The three NF-κB sites of the promoter have been shown to be important for CD40L induction in M12.4.1 cells (35). To attempt to determine the relative importance of the NF-κB and ATF binding sites for CD40L and PMA induction and promoter expression, we compared the effect of mutations of these sites in reporter assays in M12.4 and A20.3 cells. Fig. 2,A shows the mutations tested and Fig. 7 presents the results of the reporter assays. The data demonstrate that mutation of all three κB sites eliminates CD40L induction in M12.4.1 cells (Fig. 7,A) and IL-4 induction in A20.3 cells (Fig. 7 B) and greatly reduces the overall level of expression. The combination of the bZip mutation and the three κB site mutations further reduces activity by about 2-fold in M12.4.1 cells. It is clear that the κB sites of the γ1 promoter are essential for both overall activity and for CD40L and IL-4 induction, whereas the bZip site is somewhat less important. Because ATF2 is inducible in splenic B cells, it is likely that the bZip site also contributes to induction by CD40L, LPS, and B cell receptor cross-linking in splenic B cells.

Several reports have shown that C/EBPβ interacts with NF-κB functionally and physically (56, 57, 58). Stein and Baldwin (57) reported that a C/EBPβ-RelA/p50 complex activates transcription via a C/EBP binding site, but inhibits transcription from a κB site. Because of these results and because overexpression of C/EBPβ and C/EBPγ inhibit the γ1 promoter containing the bZip mutation (Fig. 6,C), we hypothesized that C/EBPβ and C/EBPγ may be able to inhibit γ1 promoter activity by interfering with NF-κB activity. To examine this possibility, M12.4.1 cells were cotransfected with NF-κBp50 and RelA along with varying doses of C/EBPβ, or with one dose of C/EBPγ, and reporter gene expression was assayed. Consistent with a previous report (35) and as shown in Fig. 8,A, overexpression of NF-κB in unstimulated cells induces the γ1 promoter by 75-fold. Addition of graded doses of an expression plasmid for C/EBPβ greatly reduces promoter activity and at a 1:1 ratio with NF-κB inhibits activity by 93%. However, C/EBPγ has only a small inhibitory effect. When the same experiment was performed with the mbZip reporter, C/EBPβ again completely inhibited activity, and C/EBPγ was only partially inhibitory at the maximum dose tested (Fig. 8 A). These data indicate that the inhibitory effect of these proteins does not require the bZip-binding site. To attempt to test whether the κB sites are required for the inhibition, the three κB sites were mutated. However, the activity of the m3κB promoter was too low to quantitate inhibition by C/EBPβ. Altogether, these data suggest that although C/EBPβ can bind the bZip element in EMSAs, it does not appear to function at this site. Instead, it may inhibit the GL γ1 promoter by interaction with NF-κB. However, C/EBPγ, which also inhibits promoter activity, does not appear to do so by inhibiting activation by NF-κB.

To further explore the mechanism of inhibition of the promoter by C/EBPβ, we tested the effect of transfection of an expression plasmid for the bZip domain of C/EBPβ. The bZip domain binds DNA and can interact with NF-κB in association with DNA and also in the absence of DNA but lacks a transactivation domain (57, 58, 59). The data in Fig. 6,D show that the C/EBPβbZip domain does not activate transcription from the multimerized C/EBP site reporter. Thus, if the bZip domain inhibits NF-κβ-induced promoter activity, this would further suggest that C/EBPβ is inhibiting promoter activity by interacting with NF-κB, rather than inducing transcription of an inhibitory factor. As shown in Fig. 8 B, the bZip domain of C/EBPβ inhibits the γ1 Luc reporter at least as well as the full-length C/EBPβ protein. As is true for the full-length C/EBPβ, inhibition does not depend on the b-Zip site in the promoter. These data are consistent with the hypothesis that C/EBPβ inhibits the GL γ1 promoter by inhibiting the activity of NF-κB.

In conclusion, the GL ε and γ1 promoters share several transcription factor binding sites, namely, a Stat6 site, a bZip site, and two (ε promoter) or three (γ1 promoter) NF-κB sites. The NF-κB binding sites of both promoters were shown previously to be essential for expression and for CD40 inducibility in M12.4.1 cells. We now demonstrate that the affinity of Stat6 for the γ1 and ε promoters differs. In addition, different transcription factors bind the bZip elements adjacent to the Stat6 sites in the two promoters. By exchanging DNA sequences between these two promoters, we determined that the differences in Stat6 binding affinity and bZip binding proteins appear to explain why the ε promoter is much more responsive to IL-4 and the combination of IL-4 and CD40L stimulation than is the γ1 promoter.

Switching to IgG1 and IgE are differentially regulated, although both isotypes are induced by IL-4 and are T dependent. Previous reports indicate that the mouse GL γ1 and ε transcripts, which are required to direct class switch recombination to a particular S region and thereby a particular C_H gene, are differentially regulated (25, 27, 28). In this manuscript, we attempt to dissect the mechanism for this differential regulation by comparing the regulation of the promoters for GL γ1 and ε transcripts. Promoter constructs consisting of elements of the IL-4-responsive region of the ε promoter substituted for the comparable region of the γ1 promoter and vice-versa were assayed in transient reporter gene experiments. Our results demonstrate that the IL-4-responsive elements within the promoters for GL γ1 and ε differ so that the ε IL-4-responsive element is more inducible by IL-4 and the combination of CD40L and IL-4, whereas the other regions of the γ1 promoter have greater activity. Experiments in which we altered the positions of the Stat6 and bZip elements in the GL γ1 promoter to mimic their positions in the GL ε promoter demonstrated that the elements themselves rather than their different positions within the promoters account for the differences in transcriptional regulation.

Both promoters contain a binding site for Stat6, a transcription factor that is activated by Jak kinases in response to IL-4 signaling (60, 61). However, the relative affinities of the ε and the γ1 Stat6 sites are ∼10:1, as assayed by competition EMSA. Adjacent to or overlapping with the Stat6 sites in both promoters is a binding site for a bZip protein. In the GL ε promoter, this site binds AP-1 transcription factors, specifically JunD and FosB in nuclear extracts from activated splenic B cells (32). Mutation of the AP-1 site eliminates both IL-4 and CD40L inducibility and basal expression (32, 36, 38). AP-1 DNA binding activity is inducible in splenic B cells by CD40L, LPS, or by cross-linking surface IgG (32, 62, 63).

The bZip site of the γ1 promoter matches a consensus site for C/EBP(NF-IL-6), T(T/G)NNGNAA(T/G) (64), and also matches the optimal binding site for ATF2 (TGACGT(A/C)A), except at position 4 (65, 66). Different members of the ATF/CREB family have different preferred binding sites, as shown by selection of sites from random pools of oligonucleotides (67). Although splenic B and M12.4.1 cells have other ATF proteins (Ref. 54 ; Fig. 4 C), ATF-2 binds preferentially at the γ1 site.

Unlike the AP-1 site in the ε promoter, which is essential for induction and for any expression of the promoter, the γ1 ATF2 site is not essential for induction of the promoter by a combination of CD40L and IL-4 in M12.4.1 cells or by PMA and IL-4 in A20.3 cells (Fig. 2, A and B). However, the ATF2/C/EBP site is important for overall activity of the promoter, because mutation of the element caused a 2- or 5-fold reduction in activity in M12.4.1 or A20.3 cells, respectively. ATF2 binding is constitutive in these B cell lines, which may explain why its binding site does not contribute to promoter induction in transient reporter assays. It is possible that because ATF2 is inducible in splenic B cells by CD40L, LPS, or by anti-δ dextran, in vivo it may be important for induction of the promoter. By contrast, it should be noted that mutation of the AP-1 site has a profound effect on the GL ε promoter, even in cells in which very low levels of AP-1 binding activity is detected (32). Thus, it is clear that the ε promoter depends much more on the AP-1 site than the γ1 promoter depends on the ATF2 site.

Experiments in which the ε(Stat6 + AP-1) sites are substituted for the comparable γ1(Stat6 + ATF2) sites resulted in a promoter that is highly inducible and is much more active than the wild-type GL ε promoter. In addition to demonstrating that the Stat6 + AP-1 sites of the GL ε promoter confer more inducibility than the γ1(Stat6 + ATF2) elements, these data demonstrate that sequences in the γ1 promoter flanking the Stat6/ATF2 sites are much more important for activity of the GL γ1 promoter than are comparable regions of the ε promoter. The latter conclusion is consistent with previous reporter gene experiments in which the effects of linker scanning mutations across the γ1 and ε promoters were examined. It was found that several different mutations within the γ1 promoter, both 5′ and 3′ to the IL-4-responsive region greatly decrease activity and inducibility (33). By contrast, linker scanning mutations across the mouse GL ε promoter revealed only two essential elements (NF-κB sites) and two elements in which mutation partially reduced activity (a B cell-specific activator protein/Pax5 site and an undefined site; Ref. 36, 38 , and 68). The high degree of dependence of the GL ε promoter on IL-4 for activity should be adaptive in preventing promiscuous production of IgE. The very low levels of IgE in serum and the low numbers of IgE⁺ cells in vivo are consistent with this stringent regulation of GL ε transcripts. Furthermore, the more promiscuous expression of IgG1 appears adaptive because this Ab can interact with several effector mechanisms to control and eliminate pathogens.

It is unclear whether the fact that AP-1 is more transiently induced than ATF2 is important for the differential regulation of the ε and γ1 promoters. Although both AP-1 and ATF2 are activated in splenic B cells treated with CD40L, LPS, or anti-Ig, AP-1 activity is expressed much more transiently. AP-1 binding activity is induced by 2 h of treatment with CD40L, is maximal at 4 h, and returns to basal levels between 24 and 48 h (32). By contrast, ATF2 binding activity remains elevated for at least 96 h in splenic B cells activated by LPS or anti-IgD (54). Although one could hypothesize that maintenance of ATF2 activity would help to maintain GL γ1 transcription over a longer time period than ε transcription, near maximal levels of both transcripts are sustained for 3–4 days in splenic B cells treated with LPS+IL-4 in culture (27, 69, 70, 71).

It is possible that AP-1 only needs to be transiently induced because it attracts the histone acetyl transferase, p300/CBP (72, 73), which is responsible for remodeling nucleosomes and which may remain bound to the GL ε promoter after AP-1 levels are reduced. By contrast, although ATF2 also binds p300 (74), it itself has histone acetyl transferase activity (75). Perhaps its binding must be maintained for optimal active chromatin structure.

Signaling via the Ag receptor or via CD40 in human B cells activates all three mitogen-activated protein kinase pathways, i.e., c-Jun N-terminal kinase, extracellular signal-regulated kinase, and p38 mitogen-activated kinase kinases (66, 76, 77, 78, 79), and ATF2 and AP-1 are differentially activated by these pathways. Signaling via the Ag receptor or by CD40 induces both phosphorylation and transcriptional activity of both ATF2 and c-Jun (77). Therefore, the large increase in DNA binding activity of ATF2 that we observed in nuclear extracts from activated splenic B cells may be attributable to induced phosphorylation (67). It is also possible, and not mutually exclusive, that binding activity is regulated by interaction with other proteins, e.g., Rb (80) and p300 (74).

C/EBPβ or both C/EBPβ and C/EBPγ inhibit transcription from the mouse GL ε or γ1 promoters, respectively, but in neither case does inhibition depend on binding to the bZip site of the promoter (Ref. 32 ; Fig. 6 C; C. H. Shen and J.S., unpublished data). C/EBPβ has been shown to inhibit transcription from κB sites and to inhibit the ability of NF-κB to bind its cognate site (57, 59). Because NF-κB is essential for activity of the GL γ1 promoter and for activation of the promoter by CD40 signaling (31, 35), we tested the effect of overexpression of C/EBPβ on activation of the GL γ1 promoter by NF-κB RelA/p50. Consistent with the results reported by Stein and Baldwin (57, 59), we observed that C/EBPβ and the C/EBPβbZip domain each inhibit promoter activation by NF-κB RelA/p50. Although the mechanism of inhibition of NF-κB activity by C/EBPβ has not been determined, two possibilities were suggested (59). Because C/EBPβ and NF-κB can bind each other in the absence of DNA, C/EBPβ might be able to squelch the NF-κB activity. An alternative is that the heterodimer has a lower affinity for the κB site than NF-κB itself. Our data do not allow us to determine whether either or both of these mechanisms are occurring.

Support for the physiological role of C/EBPβ as an inhibitor of GL γ1 transcription comes from the finding that mice with a targeted deletion of the C/EBPβ gene show increased levels of B cells bearing surface IgG1 (81). An effect on cell surface IgE expression was not reported. C/EBPβ expression is induced by IL-6 and C/EBPβ in turn activates transcription from the IL-6 gene. IL-6 induces terminal B cell differentiation toward Ab secretion and is found to increase in levels as B cells mature to the plasma cell stage (42, 82). Interestingly, plasma cells do not undergo Ab class switch recombination.

However, the mechanism of inhibition of promoter activity by C/EBPγ appears to differ, because unlike C/EBPβ, overexpression of C/EBPγ does not inhibit induction by CD40L (Fig. 6 C). Furthermore, it has never been reported that C/EBPγ interacts with NF-κB. Although overexpressed C/EBPγ has been shown to inhibit transcription in mouse B cell lines by competing with C/EBPβ for binding to DNA (55), it is difficult to understand how competition with C/EBPβ would inhibit the GL γ1 promoter. It is clear that there is much left to be understood about C/EBPγ.

One of the two signaling pathways whereby IL-4 exerts its effects results in activation of Stat6, which binds to the promoters of several IL-4-inducible genes (reviewed in Refs. 53 and 83). The IL-4-responsive regions from several genes activated by Stat6 have been studied, and these promoters show several similarities to the mouse GL γ1 and ε promoters. For example, the human GL ε and γ3 promoters and mouse CD23 promoters all contain Stat6 and NF-κB binding sites that synergize to induce transcription (84, 85, 86, 87, 88). In addition, the human GL ε promoter contains a C/EBPβ site adjacent to the Stat6 element, which is required for transcription of the promoter, and in this case, C/EBPβ activates transcription via this site (89). Although a C/EBP site has not been identified in the human γ3 promoter, cotransfection of an expression plasmid for C/EBPγ, but not C/EBPβ, synergistically induced transcription with Stat6 and NF-κB (88). The available data suggest that Stat6 does not by itself interact with the basal transcriptional machinery, but instead cooperates with other factors, including members of the bZip and NF-κB families, to effectively induce transcription in response to IL-4.

We thank Drs. Phyllis LuValle (University of Calgary) for ATF2 expression plasmids, Cathy Cooper (University of Massachusetts Medical School) for C/EBP expression and reporter plasmids, Cliff Snapper (Uniformed Services University of the Health Sciences MD) for anti-δ dextran, and William E. Paul (National Institutes of Health) for supernatant containing recombinant baculovirus mouse IL-4, and Sean Bradley and Carol Schrader, (University of Massachusetts Medical School) for helpful advice and reagents.