Induction of germline (GL) ε transcripts, an essential step preceding Ig isotype switching to IgE, requires activation of transcription factors by IL-4 and a B cell activator, e.g., CD40 ligand or LPS. We demonstrate that AP-1 (Fos and Jun), induced transiently by CD40 ligand or LPS, binds a DNA element in the mouse GL ε promoter. AP-1 synergizes with Stat6 to activate both the intact GL ε promoter and a minimal heterologous promoter driven by the AP-1 and Stat6 sites of the mouse GL ε promoter. By contrast, C/EBPβ, which trans-activates the human GL ε promoter, inhibits IL-4 induction of the mouse promoter, probably by attenuating the synergistic interaction between AP-1 and Stat6. Furthermore, AP-1 does not trans-activate the human GL ε promoter. Thus, induction of GL ε transcripts in mice and humans may be regulated differently. In addition, although mouse GL ε transcripts have a half-life of ∼100 min, the RNA level continues to increase for up to 24 h, and the promoter appears to be active for at least 2 days after B cell activation. Altogether, these data suggest that induction of AP-1 activity, although transient, is required for activation of the mouse GL ε promoter by IL-4-induced Stat6.
During Ag stimulation, resting B cells are activated and switch from expressing IgM to other Ab isotypes through the mechanism of switch recombination. At the DNA level, the switch region, located upstream of the Cμ heavy chain gene, recombines with the switch region of a downstream heavy chain gene, deleting the intervening DNA. After switching, the mature heavy chain mRNA is transcribed from the original rearranged VH gene and the recombined CH gene (reviewed in Refs. 1 and 2).
An essential step in the process of switch recombination is the induction of specific classes of germline (GL)4 transcripts, which occurs before DNA recombination. Gene targeting experiments indicate that switching requires the synthesis of GL transcripts from the CH gene (3, 4). GL transcripts from different CH genes have a similar structure, and their transcription is driven by promoters located upstream of each switch region. The transcripts initiate at a 5′-exon (or I exon) and are transcribed through the switch region, terminating at the 3′-end of the CH gene. After splicing, the germline transcripts contain the I exon and the exons of the CH gene, deleting the switch region sequences and introns of the CH gene (1, 2, 5, 6).
Specific cytokines direct isotype-specific switching by regulating the activities of GL promoters. IFN-γ, IL-4, and TGF-β selectively direct isotype switching in mouse B cells to IgG2a, to IgG1 and IgE, and to IgA and IgG2b, respectively. In addition to cytokines, a second signal that activates B cells, such as CD40 ligand (CD40L), cross-linking of B cell receptors, or stimulation of murine B cells by the mitogen LPS, is required for optimal induction of GL transcripts in splenic B cells (1, 2, 7, 8).
Several reports have shown that IL-4 and CD40L costimulation induces GL ε transcripts in both mouse and human B cells (9, 10, 11, 12). IL-4 induces phosphorylation and homodimerization of the transcription factor Stat6 through Janus kinase/STAT signaling pathways, and the dimerized Stat6 is translocated from the cytoplasm into the nucleus, where it binds the Stat6 site TTCNNNNGAA in the GL ε promoter (13, 14, 15, 16, 17). CD40L induces the transcription factors NF-κB, AP-1, and NF-AT in B cells (18). Two κB sites downstream of the Stat6 site in the GL ε promoter bind NF-κB and are responsible for promoter induction by CD40L (12, 19). Previous studies have focused on the trans-activating roles of Stat6 and NF-κB in activation of the GL ε promoter to explain the synergism between IL-4 and CD40L stimulation (12, 20). Although activation of Stat6 and NF-κB correlate with induction of GL ε transcripts, a DNA sequence element located immediately upstream of the Stat6 site in both mouse and human GL ε promoters is also required for IL-4 induction of the promoter (12, 16, 21).
This essential DNA element was previously identified as a C/EBP binding site (16). Several transcription factors of the C/EBP family have been identified so far in different tissues. They form homodimers or heterodimers that recognize the same DNA binding motif (reviewed in Ref. 22). Among these factors, C/EBPβ and C/EBPγ have been found in mouse splenic B cells, generally C/EBPβ exerting positive effects and C/EBPγ exerting negative effects on regulating promoter activities (23). It was previously found that recombinant mouse C/EBPγ and recombinant human C/EBPβ can bind to the mouse GL ε −124/−98 and human GL ε −116/−87 DNA fragments, respectively (16, 24). In addition, overexpression of human C/EBPβ and Stat6 can synergistically activate a reporter gene driven by both the C/EBP and Stat6 sites of the human GL ε promoter in IL-4-treated human HEK293 cells or BJAB cells (24). However, despite these in vitro demonstrations, no evidence has shown that C/EBP proteins at a physiologic concentration, constitutively or inducibly, are able to bind to the C/EBP site of either the mouse or human GL ε promoter, raising doubts about the physiologic roles of C/EBP proteins in induction of GL ε transcripts in normal B cells.
The current study was therefore designed to identify the essential factors binding to the C/EBP site in normal mouse splenic B cells stimulated with IL-4 and CD40L and to determine the function of these factors in induction of GL ε transcripts. We found that AP-1 transcription factors (Fos and Jun proteins) (25) are induced by CD40L and bind to the C/EBP site of the mouse GL ε promoter. In addition, induction of AP-1 activity correlates with induction of GL ε transcripts in mouse splenic B cells. Therefore, we redefine the C/EBP site as an AP-1 site.
We find that AP-1 proteins synergize with Stat6 to activate a reporter gene containing either the intact mouse GL ε promoter or containing a heterologous promoter and an enhancer consisting of the AP-1 and the Stat6 sites of the ε promoter. Conversely, C/EBPβ does not synergize with Stat6 to activate such reporter genes, and instead abolishes promoter induction by IL-4. By contrast, the C/EBP site in the human GL ε promoter allows C/EBPβ, but not AP-1, to synergize with Stat6 to activate transcription. Therefore, it appears that the highly conserved GL ε promoter sequences diverge at a critical AP-1 site in mouse and C/EBP site in human. It is possible that the diverged mouse and human GL ε promoters are regulated differently, because AP-1 and C/EBP transcription factors are subject to different signaling pathways, and because each factor can regulate the function of the other factor in induction of GL ε transcripts. Finally, a kinetics study indicates that the level of mouse GL ε transcripts continues to increase up to 24 h and is sustained for at least another 24 h post IL-4/CD40L stimulation, although the ε RNA is short-lived. These results suggest that the GL ε promoter is constantly active during this time, although AP-1 activity is induced only transiently by CD40L.
Materials and Methods
Cells and cell cultures
Splenic B cells were isolated from BALB/c mice (8–10 wk) by depletion of T cells as described previously (26). Mouse splenic B cells and the human B cell line BJAB were cultured at 37°C in an atmosphere of 5% CO2 in RPMI 1640 medium-10% FCS (FCS). The sIgM+ mouse B lymphoma cell line I.29μ clone 22D (7, 27) was grown at 37°C in an atmosphere of 8% CO2 in RPMI 1640–20% FCS.
Recombinant IL-4, CD40L, and LPS
Recombinant mouse IL-4 (a gift of W. E. Paul, National Institutes of Health, Bethesda, MD) was a culture supernatant prepared from insect cells infected with recombinant baculoviruses (28). Recombinant mouse IL-4 was added to splenic B cell cultures or I.29μ B cell cultures at 4000 U/ml. Recombinant human IL-4 (Intergen, Purchase, NY) was added to BJAB cell cultures at 10 ng/ml. CD40L is a soluble CD40L-CD8α fusion protein secreted from transfected J558L mouse myeloma cells cultured in RPMI 1640 medium-10% FCS (29) and was added to splenic B cell cultures at a final concentration of 20%. Supernatant from nontransfected J558L cell cultures was used as a control. LPS (055:B5; Sigma, St. Louis, MO), dissolved in RPMI 1640, was added to I.29μ B cell cultures at 50 μg/ml.
Sequences of oligonucleotides
The following oligonucleotide sequences (top strand) were used for DNA probes or competitor fragments in EMSAs. GL ε −124/−105: 5′-TGCCTTAGTCAACTTCCCAA-3′; octamer-binding protein (Oct): 5′-TGTCGAATGCAAATCACTAGAA-3′; Stat6: 5′-ACTTCCCAAGAACA-3′; κB: 5′-AAGGGAACTTCCAA-3′; AP-1: 5′-AGCTTGGTGACTCATCCG-3′; Ets: 5′-GGGCTGCTTGAGGAAGTATAAGAA-3′; C/EBP: 5′-TGCAGATTGCGCAATCTGCA-3′; wt (AP-1): same as GL ε −124/−105; mutated (mt) AP-1 −115: 5′-TGCCTTAGTtAACTTCCCAA-3′; mt AP-1 −117/−116: 5′-TGCCTTAcaCAACTTCCCAA-3′; wt-consensus AP-1: 5′-TGCCTgAGTCAACTTCCCAA-3′; wt-C/G (AP-1): 5′-cGCgTTAGTCAgCTTCCgcg-3′; wt(Stat6): same as Stat6; mt Stat6 −104/−102: 5′ACTTCCCAActtCA-3′; wt-C/G(Stat6): 5′-gCTTCCgcgGAAgc-3′. The bold lower case characters indicate mt base pairs, and the AP-1 and Stat6 sequences are underlined.
The luciferase reporter plasmid p(−56)FL was generated by cloning the SalI/BamHI fragment containing the minimal c-fos promoter from (−56)-c-fos-CAT (30) into the SalI/BglII sites of pXP2 vector (31). The following wild-type (wt) or mt direct-repeat nucleotide sequences derived from the mouse GL ε promoter-−124/−99 segment with an additional 5′ BamHI site and an additional 3′ HindIII site were cloned into the BamHI/HindIII sites of p(−56)FL. wt: 5′-(TGCCTTAGTCAACTTCCCAAGAACAG)2-3′; mt AP-1 −115: 5′-(TGCCTTAGTtAACTTCCCAAGAACAG)2-3′; mt AP-1 −117/−116: 5′-(TGCCTTAcaCAACTTCCCAAGAACAG)2-3′; mt Stat6 −104/−102: 5′-(TGCCTTAGTCAACTTCCCAActtCAG)2-3′; wt-consensus AP-1: 5′-(TGCCTgAGTCAACTTCCCAAGAACAG)2-3′; wt-C/G: 5′-(cGCgTTAGTCAgCTTCCgcgGAAgcG)2-3′. The underlined sequences represent the AP-1 (the 5′ side) and the Stat6 (the 3′ side) sites of the mouse GL ε promoter, respectively. The following sequences derived from the human GL ε promoter −116/−87 segment were constructed in the same manner as described for the mouse GL ε sequences. wt: 5′-(GCTGTTGCTCAATCGACTTCCCAAGAACAG)2-3′; mt C/EBP −112/−111: 5′-(GCTGaaGCTCAATCGACTTCCCAAGAACAG)2-3′; mt Stat6 −92/−90: 5′-(GCTGTTGCTCAATCGACTTCCCAActtCAG)2-3′. The underlined sequences represent the C/EBP (the 5′ side)- and Stat6 (the 3′ side)-binding sites of the human GL ε promoter, respectively. The nucleotide sequences typed with bold lower case characters are mutated from their wt sequences. The luciferase reporter plasmids containing the whole mouse GL ε promoter with the wt (−162-Luc) or mt AP-1 site (mt13-Luc) have been described previously (16). The luciferase reporter plasmids containing a minimal alkaline phosphatase promoter linked upstream with (E4APLuc) or without (APLuc) four copies of a C/EBP binding site were provided by C. Cooper, University of Massachusetts Medical Center (Worcester, MA) (23).
32) was first cloned into the EcoRI site of pBluescript(SK−) (Stratagene, La Jolla, CA), followed by cloning the BamHI/XhoI fragment containing the c-Fos cDNA from the derived plasmid into the BamHI/XhoI sites of pcDNA3. The XmnI/XbaI fragment containing the mouse FosB cDNA from FosB-pCMV2 (provided by P. Dobner) (32) was cloned into the EcoRI/XbaI sites of pcDNA3. The EcoRI/ScaI fragment containing the mouse c-Jun cDNA from c-Jun-pCMV2 (provided by P. Dobner) (32) was cloned into the EcoRI/EcoRV sites of pcDNA3. The EcoRI/XhoI fragment containing the mouse JunB cDNA from JunB-pGEM2 (ATCC 63025) was cloned into the EcoRI/XhoI sites of pcDNA3. The BamHI/XhoI fragment containing the mouse JunD cDNA from JunD-pBluescript (SK−) (ATCC 63024) was cloned into the BamHI/XhoI sites of pcDNA3. The EcoRI/XhoI fragment containing the mouse C/EBPβ cDNA from MSV/EBPβ-pBluescript (provided by S. Mcknight, Tularik, South San Francisco, CA) was cloned into the EcoRI/XhoI sites of pcDNA3. The NotI/AvrII fragment containing the human genomic c-fos sequence from pc-fos(human)-1 (ATCC 41042) was cloned into the NotI/XbaI sites of pcDNA3. The BamHI/BglII (blunt) fragment containing the human JunD cDNA from pRSV-hjD (ATCC 95655) was cloned into the BamHI/XhoI (blunt) sites of pcDNA3. The EcoRI/XhoI fragment containing the human C/EBPβ cDNA from pBlue 610 (provided by S. Akira, Osaka University) was cloned into the EcoRI/XhoI sites of pcDNA3. To generate the dominant negative c-Jun expression vector, the cDNA sequence encoding the amino acid residues from 251 to 277 (containing the DNA binding domain) of mouse c-Jun was deleted from the mouse c-Jun-pcDNA3 by oligonucleotide-directed mutagenesis, and the EcoRI/XbaI fragment from the subsequently derived plasmid was cloned into the EcoRI/XbaI sites of pcDEF3 (33).EcoRI fragment containing the mouse c-Fos cDNA from c-Fos-pCMV2 (provided by P. Dobner, University of Massachusetts Medical School) (
Transient transfection assays
All of the plasmids used for transfection were prepared by CsCl gradient purification. Mouse I.29μ B cells or human BJAB cells (5 × 107) mixed with 40 μg reporter plasmids and the indicated amount of expression plasmids in 1 ml RPMI 1640 were electroporated at 1250 μF/300 V and aliquoted into two fractions, one treated with IL-4 and the other left untreated in 10 ml complete medium for 12 h. Preparations of cell lysates and luciferase assays were performed as described previously (16).
Nuclear extracts of splenic B cells and of I.29μ B cells were prepared as described previously (34), except that 1 mM Na3VO4 and 1 mM β-glycerophosphate (Sigma) were included in buffers A and C. DNA binding reactions were performed at room temperature for 30 min in 15-μl reaction volumes containing nuclear extracts of splenic B cells (1 μg) or of I.29 μ B cells (5 μg), 1 pmol 32P-end-labeled DNA probe (25 fmol for probes containing Oct, Stat6, or κB site), 2 μg poly(dI-dC), 12.5 mM HEPES (pH 7.9), 10% (v/v) glycerol, 5 mM KCl, 0.1 mM EDTA, 0.05% Nonidet P-40, and 1 mM DTT. For DNA competition experiments, a 100-fold molar excess of unlabeled competitor oligonucleotides over the labeled probe was included in the binding reaction. For Ab supershift experiments, 1 μl of either control or the indicated specific Ab was added to the binding reaction, in which glycerol was replaced with 8% sucrose. The binding reaction mixtures were then electrophoresed in 4 or 5% native polyacrylamide gels with 0.5× Tris-borate-EDTA buffer, followed by autoradiography. Abs against mouse c-Fos (sc-52X), FosB (sc-48X), c-Jun (sc-45X), JunB (sc-46X), JunD (sc-74X), and C/EBPβ (sc-150X) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).
Total RNA from cultured splenic B cells (5 × 106) was isolated using the Ultraspec RNA isolation system (Biotecx Laboratories, Houston, TX). For reverse transcription (RT) of RNA, a 10-μl volume of annealing reaction containing total RNA (1 μg) and specific primers (50 pmol each) for GL ε transcripts (5′-CGTTGAATGATGGAGGAT-3′; Ref. 11) and for HPRT transcripts (5′-TACTAGGCAGATGGCCACAGGACTA-3′, positions 839–815, GenBank accession number J00423) was incubated at 65°C for 10 min and cooled to room temperature, followed by addition of a 10-μl volume of RT premix (1.7 mM dNTP, 6.7 U RNasin RNase inhibitor, 2× Moloney murine leukemia virus (MMLV) RT buffer, 133 U MMLV reverse transcriptase) (Promega) and by incubation at 39°C for 1 h. The RT reaction products were then diluted 16-fold with 1× MMLV RT buffer, except for samples from cells treated with IL-4 and CD40L for 24 h (Fig. 7⇓B) and for 15 h (Fig. 7⇓D), in which the RT reaction products were diluted 4-fold first and 0.5 volume of which was serially 2-fold diluted up to 5 times with 1× MMLV RT buffer to generate standard curves for measuring relative amount of cDNA. For PCR amplification of cDNA, 1 μl of the diluted RT reaction products was added to a 24-μl volume of PCR premix (1 μM each for top and bottom strand primers, 0.8 mM dNTP, 1× PCR buffer, 0.83 U HotStarTaq DNA polymerase) (Qiagen, Chatsworth, CA), and the samples were incubated at 95°C for 15 min, then at 94°C for 30 s, 50°C for 30 s, and 72°C for 30 s for 24 cycles. The top strand primers for amplifying GL ε and HPRT cDNAs are 5′-GCACAGGGGGCAGAAGAT-3′ (11) and 5′-GTTGGATACAGGCCAGACTTTGTTG-3′ (positions 601–625, GenBank accession number J00423), respectively, and the bottom strand primers are the same as those used for RT reactions. Ten microliters of the PCR product were mixed with 10 μl 2× formamide loading buffer (90% formamide and 10% 10× Tris-borate-EDTA buffer), heated at 80°C for 3 min and then chilled on ice. Fifteen microliters of each mixture were subject to 5% denaturing PAGE (7 M urea, 0.5× Tris-borate-EDTA buffer) followed by Southern blotting using 32P-end-labeled probes for detecting GL ε cDNA (5′-AGCCACTCACTTATCAGAGG-3′) (35) and HPRT cDNA (5′-CCTAAGATGAGCGCAAGTTG-3′, positions 740–759, GenBank accession number J00423). The blots were exposed to the Molecular Imager System (Bio-Rad, Richmond, CA), and the bands were quantitated by densitometry. Under the conditions described above, a single band with the expected size for GL ε transcripts (325 bp) or for HPRT transcripts (239 bp) was detected, and the intensity of the bands was in a linear range of the standard curves with a minimal regression coefficient of 0.98.
Measurement of GL ε RNA half-life
After splenic B cells were cultured in the presence of IL-4 (4000 U/ml) and CD40L (20% culture medium) for 15 h, transcription inhibitors actinomycin D (Act D, 10 μg/ml, Calbiochem, La Jolla, CA) or 5,6-dichlorobenzimidazole riboside (DRB, 30 μg/ml, Sigma) were added to the 10-ml cultures and total RNA was isolated from 5 × 106 cells at various times post inhibitor addition using the Ultraspec RNA isolation system. No significant loss of cell viability was observed by the trypan blue exclusion method (>90% viable cells). The relative amount of GL ε RNA was determined by semiquantitative RT-PCR as described above. One microgram of total RNA was electrophoresed in a 1% agarose gel followed by staining with ethidium bromide, and the relative amount of 28S rRNA in different samples were quantitated by the Fluro-S Multiimager (Bio-Rad) and used to normalize the inputs of total RNA. Linear regression of the normalized amount of GL ε RNA vs time post-inhibitor addition was plotted, and the half-life was determined as the time when 50% of GL ε RNA at time zero remains.
AP-1 transcription factors are induced by CD40L and bind to the GL ε −124/−105 promoter sequence
Previous studies have shown that the DNA sequence element TTAGTCAAC at −120/−112, relative to the first RNA initiation site and immediately upstream of a Stat6 site in the mouse GL ε promoter, is required for IL-4 induced activity of the promoter (16). Although the DNA fragment containing this sequence element was previously shown to bind recombinant C/EBPγ (16), we have been unable to demonstrate that C/EBPγ or other C/EBP proteins in nuclear extracts of B cells bind this site, suggesting that transcription factors other than C/EBP proteins might function through this site to induce GL ε transcripts. To attempt to identify factors binding to the C/EBP site of the GL ε promoter under physiologic conditions, we performed electrophoretic mobility shift assays (EMSAs) using a labeled DNA probe containing the −124/−105 DNA segment of the GL ε promoter and nuclear extracts prepared from mouse splenic B cells treated with IL-4 and/or CD40L for 4 or 24 h. As shown in Fig. 1⇓A, nuclear extracts from untreated cells form a complex with this probe (lane 2). Four hours after addition of CD40L to cells, the complex is induced by 3-fold (lane 4), but not in cells treated with IL-4 alone (lane 3). Addition of IL-4 does not further increase the complex induced by CD40L (lane 5). These results suggest that the proteins binding to the GL ε −124/−105 sequence are constitutively expressed in resting B cells and can be further induced by CD40L.
At 24 h, the amount of the complex formed with each extract, regardless of the treatment, is similar to that found with extracts from untreated cells, suggesting that the proteins involved in the CD40L-inducible complex are induced transiently on CD40L stimulation (Fig. 1⇑A, lanes 6–8). The efficacy and specificity of the IL-4 and CD40L used in the experiments were demonstrated by their ability to activate Stat6 and NF-κB binding in EMSAs, respectively (Fig. 1⇑A). In addition, the differences in the amounts of the complexes among samples are not due to differential sample loading, because nearly equivalent amounts of noninducible transcription factor Oct-1 were present in different samples (Fig. 1⇑A).
Because the sequence of the −120/−114 segment closely resembles a consensus site for AP-1 (TGAGTCA, differing only at the second position), we examined whether AP-1 proteins are in the binding complex. As shown in Fig. 1⇑B, addition of an unlabeled double-stranded oligonucleotide containing the consensus AP-1 sequence as a competitor in EMSAs completely abolished this complex, whereas addition of a 100-fold molar excess of a consensus binding site for different transcription factors (C/EBP or Ets) did not or only partially reduced the complex. These results suggest that this complex contains AP-1 proteins.
To identify which members of the AP-1 family form the complex, we performed Ab supershift assays (Fig. 1⇑C). Components of the complex formed with nuclear extracts from splenic B cells induced for 4 h with IL-4 and CD40L were supershifted when specific Abs against c-Fos, FosB, c-Jun, or JunD were added, whereas components of the complex formed with extracts from cells induced for 24 h were supershifted with FosB, c-Jun, and JunD Abs but not with c-Fos Abs. No JunB was detected in the complexes formed at 4 or 24 h (even in overexposed autoradiographs). An Ab that can specifically and efficiently supershift C/EBPβ-containing complexes in EMSAs did not supershift the complex (lanes 9 and 17), indicating that components of the complex are shifted specifically by Abs against certain AP-1 proteins. In addition, none of the Abs used in these experiments shifted the labeled probe if no nuclear extracts were included in the binding reactions. These Abs have been tested and shown to be able to supershift their target proteins in EMSAs using nuclear extracts from cells that express all these AP-1 proteins (data not shown). These results suggest that c-Fos, FosB, c-Jun, and JunD are present in the CD40L-inducible AP-1 complexes at 4 h, and FosB, c-Jun, and JunD are components of the AP-1 complex(es) at 24 h. Due to the limitation of our detection method, we could not determine whether c-Fos is still present in the AP-1 complex at 24 h. We have used competition and Ab supershift EMSAs to test for the presence of two other families of transcription factors, cAMP response element binding protein/activating transcription factor and C/EBP, which bind to sequences closely related to an AP-1 site. These factors were not found in this AP-1 complex (Fig. 1⇑, B and C, and data not shown). Therefore, AP-1 proteins appear to be the only proteins in this complex.
Both the AP-1 and Stat6 sites of the GL ε promoter are necessary and together are sufficient to confer IL-4 inducibility
Induction of GL ε transcripts in B cells by IL-4 requires a second signal such as CD40L or LPS which activates NF-κB and AP-1 transcription factors (Refs. 7, 8, 10, 18 and 36 and data not shown). The requirement for IL-4 and CD40L costimulation to induce GL ε transcripts in B cells has been suggested to be due to synergistic activation of the promoter by Stat6 and NF-κB (12, 20). It has been shown, however, that even in the presence of NF-κB, the GL ε promoter cannot be activated by IL-4 treatment if the sequence corresponding to the AP-1 site identified above is mutated (16), suggesting that activation of the GL ε promoter by IL-4-induced Stat6 is also dependent on the AP-1 activities induced by CD40L or LPS.
A previous report demonstrated that a DNA segment containing the −124/−98 sequence of the mouse GL ε promoter is sufficient to confer IL-4 inducibility to a heterologous promoter and that 8-nucleotide linker sequences replacing the newly identified AP-1 site or the Stat6 site within this segment eliminate IL-4 inducibility of the promoter (16). However, due to the substantial sequence replacement, it was not clear from the previous experiments whether the AP-1 site and the Stat6 site are both necessary, nor was it certain whether they are sufficient for IL-4 inducibility of the reporter plasmids.
To address these questions, more precise and limited nucleotide substitutions (one to three nucleotides) were introduced into the reporter constructs with identically oriented dimers of the −124/−99 segment containing the AP-1 and the Stat6 sites of the GL ε promoter. The plasmids were transiently transfected into I.29μ B lymphoma cells. These cells express GL ε transcripts which can be further induced by IL-4 treatment (7, 16), and they constitutively express a low level of AP-1 (which includes at least c-Fos, FosB, c-Jun, JunB, and JunD, identified by Western blotting in data not shown) and acquire Stat6 activity on IL-4 stimulation (20). Presumably because of the constitutive expression of AP-1 in I.29μ cells, we found it unnecessary to induce AP-1 activity in reporter assays. Fig. 2⇓A shows that the transcriptional activity of the wt reporter plasmid (wt) is induced 18-fold by IL-4 stimulation, whereas mutation of the AP-1 site (mt AP −115 or mt AP −117/−116) or the Stat6 site (mt Stat6 −104/−102) in the reporter plasmid completely abolishes promoter activity, demonstrating that both AP-1 and Stat6 sites are necessary for IL-4-induced activity. To further demonstrate that AP-1 is able to trans-activate the GL ε segment, the T at position −119 was replaced with a G so that a consensus AP-1 site was created in the enhancer sequence, resulting in an enhancement of the reporter activity to 125-fold in the presence of IL-4 and 6-fold in the absence of IL-4.
To test whether both the AP-1 and the Stat6 binding sites are sufficient to confer IL-4 inducibility, we kept both the AP-1 (TTAGTCA) and the Stat6 (TTCNNNNGAA, underlined sequence) binding sequences unchanged but replaced all other positions with alternating CG nucleotides, based on the assumption that no DNA sequence-specific transcription factors should be able to function on both wt and the alternating CG sequences. Indeed, we found that the mt reporter plasmid (wt-C/G) was still induced by IL-4 and had an activity similar to that of the wt reporter (Fig. 2⇑A). These results suggest that no transcription factors other than those binding to the AP-1 and Stat6 sites are required for induction by IL-4. Therefore, the AP-1 and Stat6 sites together are sufficient to confer IL-4 inducibility.
To assure that the IL-4 inducibilities of the wt and the mt reporter plasmids correlate with the abilities of their corresponding AP-1 and Stat6 binding sites to bind AP-1 and Stat6, we performed EMSAs with DNA probes containing the wt and mt AP-1 or Stat6 binding sequences used in the reporter plasmids. Nuclear extracts were prepared from I.29μ B cells unstimulated (−) or stimulated (+) with LPS and IL-4. Although AP-1 is expressed constitutively at a low level in I.29μ B cells, LPS was added because it can further induce AP-1 in these cells, allowing us to visualize the AP-1 complex in EMSAs. Fig. 2⇑B shows that, regardless of the flanking nucleotide sequence, DNA fragments containing the wt or consensus AP-1 binding sequences (lanes 2, 8, and 10) bind AP-1, whereas fragments with mt AP-1 sites (lanes 4 and 6) do not. Likewise, fragments containing wt Stat6 binding sites (lanes 12 and 16) bind Stat6, whereas a fragment with a mt Stat6 site does not (lane 14). These complexes contain AP-1 and Stat6 proteins because they can each be competed specifically by consensus AP-1 and Stat6 fragments in EMSAs (data not shown). The results of EMSAs are in accordance with the results of the reporter gene assays, in which the IL-4 inducibility is manifest only in reporter plasmids containing AP-1 and Stat6 sites capable of binding their corresponding AP-1 and Stat6 proteins.
Different members of the AP-1 family synergize with Stat6 to trans-activate the reporter plasmid containing both AP-1 and Stat6 sites
To directly determine whether AP-1 factors can induce transcription from the GL ε promoter, we tested the effect of cotransfection of expression plasmids for Fos and Jun proteins on activity of the reporter plasmid containing the wt −124/−99 dimer from the GL ε promoter in I.29μ B cells. Luciferase activities were measured in transfected cells, untreated or treated with IL-4 to activate Stat6. It was not necessary to cotransfect a Stat6 expression plasmid because IL-4 induces abundant activated Stat6 in I.29μ B cells (Fig. 2⇑B, lane 12).
As shown in Fig. 3⇓A, overexpressing Fos and Jun proteins in pairs greatly increases the activity of the reporter plasmid in the presence of IL-4. These data indicate that AP-1 proteins synergize with IL-4-activated Stat6 to induce transcription. The synergy appears to be mediated by heterodimers of Fos and Jun proteins, because in the presence of IL-4, cells overexpressing either Fos proteins or Jun proteins alone gave reporter activities similar to cells without AP-1 overexpression. Synergistic activation of the promoter by AP-1 and Stat6 requires both cognate AP-1 and Stat6 sites, because mutation at either the AP-1 or the Stat6 site in the reporter plasmids abolishes promoter activity, despite the presence of activated Stat6 together with overexpressed FosB/JunD or c-Fos/JunD (Fig. 3⇓B).
Consistent with the results shown in Fig. 3⇑A, overexpressing a dominant negative c-Jun, lacking the DNA binding domain, nearly eliminates IL-4 activation of the reporter plasmid (Fig. 3⇑C). This result is presumably due to formation of defective AP-1 complexes (consisting of endogenous Fos proteins and the overexpressed dominant negative c-Jun) that are incapable of binding to DNA (37). These data further support the hypothesis that endogenous Fos and Jun proteins in I.29μ B cells synergize with Stat6 to activate the reporter plasmid containing both the AP-1 and Stat6 sites of the ε promoter.
Both c-Fos/JunD and FosB/JunD are capable of synergizing with Stat6 to activate the whole GL ε promoter
The reporter plasmid used in the above experiments contains a dimer of only the AP-1 and Stat6 sites. Because it is possible that in the context of the intact GL ε promoter the functions of AP-1 and Stat6 may be altered, we next examined whether AP-1 is able to synergize with Stat6 to activate a reporter plasmid containing the intact GL ε promoter. We cotransfected a reporter plasmid containing the −162/+53 GL ε promoter (including the first 53 nucleotides of Iε exon) along with expression plasmids for c-Fos, FosB, or JunD or their combinations in reporter gene assays in I.29μ B cells. We chose these three AP-1 proteins because they are detected in nuclear extracts from CD40L-treated splenic B cells and form complexes with the AP-1 site of the GL ε promoter in EMSAs (Fig. 1⇑C).
As shown in Fig. 4⇓A, the GL ε promoter is synergistically induced by the combination of FosB/JunD or c-Fos/JunD and IL-4-induced Stat6. In cells overexpressing FosB/JunD the GL ε promoter activity is induced 49-fold in the presence of IL-4 and 11-fold in the absence of IL-4, relative to the promoter activity in cells transfected with empty expression vectors and no IL-4 treatment (1-fold). Similarly, in cells overexpressing c-Fos/JunD proteins, promoter activity is induced 16-fold in the presence of IL-4 and 2-fold in the absence of IL-4. In the absence of cotransfected AP-1 proteins, promoter activity is induced 7-fold by IL-4. These data indicate that both FosB/JunD and c-Fos/JunD are capable of synergizing with Stat6 to activate the GL ε promoter, whereas the basal and IL-4-induced promoter activities in cells overexpressing c-Fos, FosB, or JunD alone are not significantly different from those in cells transfected with empty expression vectors. The reporter gene assays taken together with the EMSA data presented above suggest that the AP-1 proteins involved in the synergistic activation of the promoter are heterodimers of FosB/JunD and c-Fos/JunD. As expected, synergistic activation of the GL ε promoter by Stat6 and FosB/JunD or c-Fos/JunD requires the AP-1 site of the promoter, because disruption of the AP-1 site in the reporter plasmid (mt13) abolishes the synergy (Fig. 4⇓B).
Overexpression of C/EBPβ abolishes the IL-4 inducibility of the GL Cε promoter.
The DNA-binding domains of mouse C/EBPγ (38) and full-length recombinant mouse C/EBPβ specifically bind in EMSAs to the GL ε AP-1 site identified in this study (Ref. 16 and data not shown), consistent with the report that an AP-1 binding sequence is also a consensus C/EBP site (39). To determine whether C/EBPβ can also synergize with Stat6 to activate the GL ε promoter, we performed cotransfection experiments in I.29μ B cells using the reporter plasmid containing the GL ε promoter and an expression plasmid for mouse C/EBPβ. We found that overexpression of mouse C/EBPβ, in the presence or absence of IL-4, does not enhance promoter activity (Fig. 4A⇑). Interestingly, the IL-4 inducibility of the promoter was abolished when C/EBPβ was overexpressed in cells. These data indicate that mouse C/EBPβ, in contrast to AP-1 proteins, cannot synergize with Stat6 to activate the GL ε promoter, despite its potential binding ability to the AP-1 site of the promoter, and also that C/EBPβ inhibits the function of Stat6 in promoter activation.
To demonstrate that the overexpressed C/EBPβ we used is functional, we performed reporter assays in I.29μ B cells using the same mouse C/EBPβ expression vector and a reporter gene driven by four copies of Ig μ intron C/EBP binding sites as the enhancer sequence (23). As shown in Fig. 4⇑C, the reporter with C/EBP sites (E4APLuc) is activated by more than 5-fold when C/EBPβ is overexpressed compared with results with an empty expression vector. The overexpressed C/EBPβ cannot activate the reporter gene if it lacks C/EBP enhancer sequences (APLuc). In conclusion, these data indicate that the transfected C/EBPβ expression vector can produce functional C/EBPβ proteins which have trans-activating activity independent of IL-4 treatment in I.29μ B cells. Therefore, the inability of C/EBPβ to trans-activate the GL ε promoter is not due to production of defective C/EBPβ proteins.
Overexpression of C/EBPβ abolishes the synergistic activation of the reporter by AP-1 and Stat6
It is possible that C/EBPβ abolishes IL-4 inducibility of the GL ε promoter by interrupting the functional interaction between AP-1 proteins and Stat6. To test this possibility, we performed reporter gene assays in I.29μ B cells using the reporter plasmid driven by the GL ε AP-1 and Stat6 sites and cotransfecting varying amounts of C/EBPβ- and c-Fos/JunD-expression plasmids. Fig. 5⇓A shows that the promoter activity is increased by 2750-fold when cells were transfected with 20 μg each of c-Fos/JunD expression vectors and stimulated with IL-4, demonstrating the strong synergistic activation of the promoter by AP-1 proteins and Stat6. The synergy is decreased dramatically when 20 μg C/EBPβ-expression vector are cotransfected and abolished when 40 or 80 μg C/EBPβ expression vector are cotransfected, becoming equal to the IL-4-induced activity in the absence of cotransfected AP-1. These data indicate that C/EBPβ inhibits the synergistic induction of the promoter by AP-1 and Stat6.
Results of a reciprocal experiment (Fig. 5⇑B), in which the amount of C/EBPβ expression vector was fixed (20 μg), whereas variable amounts of c-Fos/JunD expression vectors were cotransfected (from 0 to 80 μg each), show that both the IL-4 inducibility and synergy with AP-1 which are inhibited by C/EBPβ can be restored by coexpressing excess c-Fos/JunD in cells. These data suggest that AP-1 and C/EBPβ play competitive roles in regulating the mouse GL ε promoter, AP-1 interacting with IL-4-activated Stat6 to synergistically activate the promoter, and C/EBPβ preventing AP-1 from interacting with Stat6, therefore down-regulating the GL ε promoter activity.
C/EBPβ, but not AP-1, activates the human GL Cε promoter
In contrast to the above results, it has been demonstrated that human C/EBPβ can synergize with Stat6 to activate a reporter plasmid driven by a DNA segment containing both the C/EBP and Stat6 sites from the human GL ε promoter, which is similar to but not identical with the equivalent mouse sequence used in our study (Ref. 24 and Fig. 6⇓A). We hypothesized that the differences between mouse and human may be due to the divergent sequences immediately upstream of the Stat6 site in the mouse and human GL ε promoters. In the mouse promoter, the DNA sequence from −120 to −114 appears to be a better AP-1 site than a C/EBP site, whereas in the human promoter, the DNA sequence from −113 to −105 appears to resemble an antisense consensus C/EBP site ([A/C]TTNCNN[A/C]A, differing only at the first position) rather than an AP-1 site.
To test this hypothesis, we performed cotransfection assays to compare the abilities of AP-1 and C/EBPβ transcription factors from mouse and human origins to synergize with Stat6. We used identical reporter constructs containing either the mouse −124/−99 or human −116/−87 segment of the GL ε promoter, and the assays were performed in either mouse I.29μ B cells or human BJAB cells, in accordance with the species origin of the promoters. As shown in Fig. 6⇑B, both mouse and human c-Fos/JunD synergistically activate the reporter with the mouse enhancer sequence in I.29μ B cells treated with IL-4, whereas neither mouse nor human C/EBPβ does. The opposite result was observed using reporters with the human enhancer sequence. As shown in Fig. 6⇑D, both human and mouse C/EBPβ synergistically activate the reporter with the human enhancer sequence in human BJAB cells treated with IL-4, whereas neither human nor mouse c-Fos/JunD does. The synergy shown in both reporter assays depends on both the Stat6 and the AP-1 or C/EBP sites for mouse or human, respectively, because mutation at either site eliminates this synergy (Fig. 6⇑, C and E).
The inability of C/EBPβ and c-Fos/JunD to synergize with Stat6 to activate the mouse and the human reporters in mouse I.29μ B cells and human BJAB cells, respectively, is not due to the lack of appropriate machineries to generate functional C/EBPβ in I.29μ B cells or c-Fos/JunD in BJAB cells, because in the presence of IL-4, overexpression of human C/EBPβ in mouse I.29μ B cells and overexpression of mouse c-Fos/JunD in human BJAB cells, can synergistically activate the human and mouse reporters, respectively (data not shown).
In conclusion, these data suggest that AP-1, but not C/EBPβ, is capable of synergizing with Stat6 to activate the mouse GL ε promoter, and by contrast, C/EBPβ, but not AP-1, can synergize with Stat6 to activate the human GL ε promoter. Therefore, mouse and human GL ε promoter sequences diverge at this critical AP-1 or C/EBP site, despite the fact that they share highly conserved sequences at all other positions that have been shown to be important for regulated expression of the GL ε promoters.
Kinetics of induction and half-life of GL ε transcripts in mouse splenic B cells stimulated with IL-4 and CD40L
Because our current studies together with previous findings demonstrate that IL-4-induced Stat6 and CD40L-induced AP-1 and NF-κB activate the GL ε promoter, we wished to determine whether induction of these transcription factors coincides with induction of GL ε transcription in mouse splenic B cells stimulated with IL-4 and CD40L.
As shown in Fig. 7⇓A (lanes 3, 6, 9, 12, 15, and 18), in splenic B cells the DNA-binding activities of all three factors are induced rapidly 1 h post stimulation, achieve maximal levels by 4 h, and decline with different kinetics. AP-1- and NF-κB binding activities return to near basal levels after 24 h, whereas induced Stat6 binding activity persists for at least 48 h post stimulation. Competition experiments using consensus binding sites (lanes 4, 7, 10, 13, 16, and 19) or unrelated oligonucleotides (lanes 5, 8, 11, 14, 17, and 20) indicate binding is specific for each probe. The samples for different time points were loaded equally, because similar levels of the noninducible Oct-1 protein were detected in the different samples.
Consistent with the rapid induction of AP-1, Stat6, and NF-κB binding activities, GL ε transcripts are detectable by semiquantitative RT-PCR 2 h after addition of IL-4 and CD40L (Fig. 7⇑B). No GL ε RNA could be detected at any time point if B cells were treated with either IL-4 or CD40L alone, indicating that production of GL ε RNA requires both signals and that the ε RNA detected is not from preactivated B cells in spleen. As shown in Fig. 7⇑C, the RNA level continued to increase up to 24 h, at which time the steady state level was reached.
Interestingly, achievement of maximal levels of GL ε transcripts does not appear to require maximal induced levels of AP-1 and NF-κB activities, because both AP-1 and NF-κB binding activities are maximal at 4 h and have returned to near basal levels after 24 h, whereas the level of GL ε transcripts is still low at 4 h and only reaches maximum at 24 h. One possible explanation for this result is that the functions of CD40L-induced AP-1 and NF-κB are to initiate transcription and to allow IL-4-induced Stat6 to bind and activate transcription through the Stat6 site on the GL ε promoter. Therefore, transient induction of AP-1 and NF-κB might be sufficient to support activation of the GL ε promoter by Stat6.
However, another possible explanation for the kinetics of induction of GL ε transcripts would be that GL ε transcripts are extremely long-lived. To address this possibility, we measured the half-life of GL ε RNA in splenic B cells after induction by IL-4 and CD40L. Transcription inhibitors, Act D or DRB, were added to the cultures after splenic B cells were treated with IL-4 and CD40L for 15 h, and the amount of RNA remaining at various times after inhibitor addition was determined by semiquantitative RT-PCR (Fig. 7⇑D). Fig. 7⇑E presents the quantitation of these data showing that after transcription is blocked by either Act D or DRB treatment, GL ε transcripts have a short half-life, ∼100 min. Because the RNA level continues to increase up to 24 h, the short half-life suggests that the GL ε promoter is constantly active for at least 2 days after addition of CD40L and IL-4. Consequently, these data support the hypothesis that transient induction of AP-1 and NF-κB may be sufficient to support transcriptional activation of the GL ε promoter by Stat6.
AP-1 is required for transcriptional activation of the Ig GL Cε promoter and C/EBPβ inhibits the promoter
In this study, we identify the transcription factor binding site which is located immediately upstream of the Stat6 site in the mouse GL ε promoter and which is required for IL-4-induced transcription by Stat6 to be an AP-1 site. Endogenous AP-1 factors in nuclear extracts from mouse splenic B cells specifically bind this site in EMSAs (Fig. 1⇑). The AP-1 site sequence TTAGTCA at −120 to −114 in the mouse GL ε promoter is identical with sequences found in the control regions of the phorbol ester-inducible genes of SV40 and polyoma viruses, which have been shown to bind AP-1 factors induced by TPA treatment (40).
Our data suggest that AP-1 proteins, but not C/EBP proteins, are the endogenous proteins driving transcriptional activation of the Ig GL ε promoter in normal mouse splenic B cells. First, AP-1 DNA binding activities are detected in EMSAs using the mouse GL ε promoter −124/−105 fragment, and their binding activities are further enhanced in nuclear extracts from mouse splenic B cells treated with CD40L or LPS (Fig. 1⇑ and data not shown). Conversely, we were unable to detect C/EBP binding to the ε promoter fragment using nuclear extracts from LPS-treated splenic B cells or from I.29μ B cells, despite the fact that C/EBPβ is induced in splenic B cells by LPS treatment (Ref. 23 and data not shown). Second, functional data indicate that AP-1 factors synergize with Stat6 to activate transcription, whereas C/EBPβ abolishes promoter activity induced by IL-4 in reporter gene assays. In addition, the sequence upstream of the Stat6 site in mouse GL ε promoter does not fit a consensus C/EBP motif, T(T/G)NNGNAA(T/G) as found in many C/EBP binding sites (41). The conserved G residue in the center of the consensus C/EBP motif is not found in either orientation of the sequence encompassing the AP-1 site of GL ε promoter. With the exception of recombinant C/EBPβ or C/EBPγ purified from transformed Escherichia coli, C/EBPβ from all sources, including in vitro-translated C/EBPβ from reticulocyte lysates and C/EBPβ in nuclear extracts from overexpressing HEK293 cells or from LPS-treated splenic B cells, failed to bind the GL ε promoter fragment −124/−105 in EMSAs (data not shown). C/EBPβ weakly binds to AP-1 sites (39), which can explain why we find that C/EBPβ binds to the ε −124/−105 fragment when sufficient amounts of proteins are present in the binding reactions. Because, at physiologic concentrations, C/EBP, in contrast to AP-1, does not detectably bind the ε −124/−105 fragment, we therefore redefine the previously defined C/EBP site in mouse GL ε promoter as an AP-1 site.
Interestingly, although C/EBPβ may not bind the AP-1 site of the mouse GL ε promoter in vivo, overexpressed C/EBPβ can down-regulate the promoter activities induced by cooperative interaction between AP-1 and Stat6. A plausible mechanism to explain down-regulation by C/EBPβ is the formation of Fos/C/EBPβ and Jun/C/EBPβ cross-family heterodimers through their leucine zipper domains (39), thereby preventing AP-1 from binding to the AP-1 site in the GL ε promoter. This is analogous to the proposed mechanism by which Fos and Jun inhibit transcriptional activation by C/EBPβ of a reporter plasmid containing C/EBP binding sites (39). Another possibility is competition between AP-1 and C/EBPβ for essential molecules involved in induction of the promoter, such as Stat6, coactivators, and the basal pol II transcription machinery. Although the mechanism is unknown, it is likely that the induction of GL ε transcripts, and therefore IgE production by mouse B cells, will be inhibited if C/EBPβ is induced. Thus, it is possible that the balance between AP-1 and C/EBPβ activities induced by different stimuli regulates GL ε transcripts.
It has been previously shown that c-Fos, FosB, Fra-1, JunB, and JunD cellular mRNA and, in some cases, proteins are induced rapidly in mouse splenic B cells treated with CD40L (42, 43). Consistent with these findings, we detect binding to the GL ε promoter −124/−105 segment by c-Fos, FosB, c-Jun, and JunD at 4 h and by FosB, c-Jun and JunD at 24 h after addition of CD40L. We did not examine Fra-1 and Fra-2 Fos proteins in our study. It is very likely that c-Fos and FosB form heterodimers with c-Jun and JunD and activate the GL ε promoter in vivo, because Fos proteins do not form homodimers but do heterodimerize with Jun proteins and because Fos/Jun heterodimers are more stable and have much higher DNA-binding affinities than Jun homodimers (25, 44). In addition, we found that coexpressed c-Fos/JunD or FosB/JunD synergize with Stat6 to activate transcription, whereas overexpressed JunD alone does not (Figs. 3⇑ and 4⇑). JunB did not bind to the GL ε promoter −124/−105 fragment in EMSAs, although JunB DNA-binding activity can be detected when a consensus AP-1 sequence is used as the probe in EMSAs (data not shown), suggesting that JunB might not bind to the GL ε promoter either as a homodimer or as a heterodimer with Fos. However, it is possible that JunB may exert an indirect effect on the GL ε promoter by altering the stoichiometry of AP-1 dimers.
Kinetics of expression of GL ε transcripts and class switching
Many reports have examined the kinetics of induction of GL ε transcripts in mouse B cells stimulated with IL-4 in combination with CD40L or LPS (5, 8, 11). In general, maximal levels of GL ε transcripts are achieved 1 day after B cells are stimulated in vitro, and this level is maintained for at least 4 days, at which time the mature, productive mRNAs are detected. The early induction events, however, have not been studied before. We detected the GL ε transcripts in mouse splenic B cells as early as 2 h after addition of IL-4 and CD40L. This suggests that all of the essential transcription factors for activation of the GL ε promoter by IL-4 and CD40L must be induced immediately after cells receive the signals. Indeed, we detect induced levels of Stat6, AP-1 and NF-κB DNA binding activities as early as 1 h after cells are treated with IL-4 and CD40L, consistent with the induction of GL ε transcripts by 2 h (Fig. 7⇑A).
Interestingly, whereas the level of GL ε transcripts is low at 4 h and maximal at 24 to 48 h, AP-1 and NF-κB binding activities are maximal at 4 h and decline to near basal levels by 24 to 48 h (Fig. 7⇑A). This raises doubts about their roles in supporting transcription as the maximal RNA level is achieved. It is possible that CD40L-induced AP-1 and NF-κB initiate transcription and allow IL-4-induced Stat6 to bind and activate the GL ε promoter. Transient induction of AP-1 and NF-κB by CD40L may be sufficient to support activation of the GL ε promoter by the more long-lasting IL-4-induced Stat6.
Resting mouse splenic B cells treated with IL-4 and CD40L can express IgE only after cells have divided at least six times; this normally requires at least 4 days (45). Due to the relatively short half-life of GL ε transcripts (100 min), prolonged active transcription from the GL ε gene is necessary to maintain the steady state level of RNA in B cells treated with IL-4 and CD40L. Several reports have demonstrated that transcription of the GL CH gene loci per se is not sufficient for switching and that the processed transcripts, the process of transcript splicing, and/or the products of spliced-out intron sequences may participate in switch recombination (46, 47, 48). Prolonged active transcription may provide a continuous mode of RNA processing throughout the whole period of switching, thereby increasing the probability of targeting switch recombination to a specific CH gene.
Synergistic induction of the Ig GL ε promoter by Stat6, AP-1, and NF-κB
As in most IL-4-inducible genes, the Stat6 binding site plays a central role for activation of GL ε transcription by IL-4. However, Stat6 appears to be unique among Stat proteins in that it is unable to activate a promoter containing only a Stat6 site even when the site is multimerized (16, 20, 49, 50). Interaction between Stat6 and other transcription factors induced by CD40L is apparently required for Stat6 to activate the GL ε promoter. Here we show that AP-1, induced by CD40L, is able to synergize with Stat6 and is essential for activation of the GL ε promoter. Physical association between AP-1 and NF-κB (51) and between Stat6 and NF-κB (20) has been previously demonstrated, and such cross-family interactions result in synergistic activation of transcription. Each of these two sets of interactions, although probably contributing to optimal activation of the promoter, is not by itself sufficient to induce GL ε transcripts. The evidence for this is that when B cells are stimulated with either CD40L or LPS, both AP-1 and NF-κB are induced, but CD40L or LPS alone does not induce GL ε transcripts, ruling out the possibility that AP-1 and NF-κB together are sufficient to activate transcription. Furthermore, a reporter gene containing the intact GL ε promoter with an AP-1 site mutation cannot be induced in IL-4-treated I.29μ B cells that contain abundant activated Stat6 and NF-κB, indicating that interaction between Stat6 and NF-κB is not sufficient (16) (Fig. 4⇑B). In this study, we provide evidence for another type of interaction, that between AP-1 and Stat6. The reporter plasmid containing only the AP-1 site and the Stat6 site cannot be induced by IL-4 in I.29μ B cells if the AP-1 site is mutated. In addition, several different combinations of Fos and Jun proteins strongly synergize with Stat6 to activate transcription, whereas expression of dominant negative c-Jun suppresses transcription, indicating that the trans-activating ability of Stat6 and consequently the IL-4 inducibility of the promoter require the interaction between AP-1 and Stat6.
The abilities of AP-1 and NF-κB to synergize with Stat6 differ, however, in that AP-1 alone is sufficient to enable Stat6 to function, whereas NF-κB alone is not. This difference indicates that CD40L-induced AP-1 and NF-κB do not function redundantly in synergizing with Stat6 to activate the GL ε promoter. Evidence suggests that the direct association of NF-κB with Stat6 may enhance Stat6 DNA binding affinity to the GL ε promoter at its adjacent Stat6 and κB sites (20). However, the physical basis for the ability of AP-1 to synergize with Stat6 is unknown. Perhaps AP-1 promotes the function of Stat6 in activating transcription by attenuating the repressor activity of BCL-6, which has recently been shown to bind the Stat6 site of the GL ε promoter and to suppress the promoter activity induced by IL-4 (52). Alternatively AP-1 could disrupt the nucleosomal structure of the GL ε promoter and facilitate subsequent binding by Stat6. This model is suggested by the finding that prior binding of AP-1 to a DNA template containing both AP-1 and SRY sites appears to disrupt the nucleosomal structure and to enhance binding of the transcription factor SRY to the DNA template (53). Either or both of the above possibilities are consistent with our proposed model.
The essential roles of Stat6 and NF-κB p50 in induction of GL ε transcripts in vivo have been demonstrated in gene knockout experiments, in which IL-4 and CD40L are unable to induce GL ε transcripts or IgE in B cells from either Stat6−/− or NF-κB p50−/− deficient mice (35, 54, 55). By contrast, GL ε transcripts can be induced by IL-4 plus CD40L in B cells from fosB−/− knockout mice or by IL-4 plus LPS in B cells from c-fos−/− knockout mice, although the steady state level of GL ε RNA appears somewhat reduced, compared with wt littermates (data not shown). It is likely that other Fos proteins may compensate for the loss of one Fos protein in these single-knockout mice. We speculate that a double- or a triple-fos knockout mouse would show a severe or complete deficiency in production of GL ε transcripts and IgE.
Regulation of Ig GL ε transcripts differs between mouse and human B cells
Despite highly conserved GL ε promoter sequences in mice and humans (Fig. 6⇑A), our data suggest that these two promoters diverge at the AP-1 site for mice and the C/EBP site for humans. We found that c-Fos/JunD, but not C/EBPβ, synergizes with Stat6 to activate transcription from mouse GL ε sequences, and conversely, C/EBPβ, but not c-Fos/JunD, synergizes with Stat6 to activate transcription from the corresponding human GL ε segment. The selective activation by AP-1 or by C/EBPβ appears to be due to different DNA-binding specificities of these two promoters, rather than differential trans-activating abilities of the mouse and human homologues of the transcription factors or unique transcription machineries in mouse or human cells. The antisense sequence of the human GL ε promoter −113/−105 fragment, 5′-TTGAGCAAC-3′, fits the consensus C/EBP site except the last base. In contrast, this sequence is unlikely to bind AP-1, because two important bases of the consensus AP-1 sequence 5′-TGAGTCA-3′ are changed. Consequently, C/EBPβ but not AP-1 synergizes with Stat6 to activate transcription with human GL ε sequences.
Both the C/EBP site in the human and the AP-1 site in the mouse GL ε promoter appear to play similar roles in promoter activation by IL-4. The reporter plasmid containing the whole human GL ε promoter with mutations in the C/EBP site 5′ to the Stat6 site cannot be induced by IL-4 in human B lymphoma cells DND39 (21), indicating that C/EBP proteins are required for IL-4 inducibility of the human GL ε promoter. Interestingly, physical association and functional synergism between C/EBPβ and NF-κB have also been previously demonstrated (56, 57). Therefore, the proposed model for induction of mouse GL ε promoter by AP-1, Stat6, and NF-κB should also apply for induction of human GL ε promoter, except the role of AP-1 may be substituted by C/EBPβ in human.
Although AP-1 and C/EBPβ may play analogous roles for induction of GL ε transcripts by IL-4, they belong to two distinct families of transcription factors and are subject to different activation pathways in cells. Therefore, the mouse and human GL ε promoters are probably not regulated identically. For example, induction of GL ε transcripts in mouse splenic B cells requires IL-4 plus CD40L or LPS, whereas induction of GL ε transcripts in human peripheral blood B cells requires only IL-4, although the response can be further enhanced by some NF-κB inducers (6). In this study, we found AP-1 activity is induced in mouse splenic B cells by CD40L, but we could not detect C/EBPβ in nuclear extracts from cells either untreated or treated with CD40L for 4 h by Western blotting (data not shown). It has not been reported, however, whether C/EBPβ can be induced by CD40L or other stimuli in normal human B cells.
In summary, we redefine the transcription factor binding site immediately upstream of the Stat6 site in mouse GL ε promoter as an AP-1 site. Activation of the mouse GL ε promoter by IL-4 depends on AP-1 proteins binding to this site. By contrast, a C/EBP site resides at the equivalent position in the human GL ε promoter. Although AP-1 and C/EBP proteins exclusively function to activate one promoter but not the other, their mutual antagonistic abilities potentially enable each of them to regulate GL ε transcripts in both mouse and human B cells. The discrepancies between the mouse AP-1 site and the human C/EBP site must be taken into consideration when analogies are drawn between the regulations of switching to IgE in mice and humans.
We thank W. E. Paul for mouse rIL-4, C. Cooper for the APLuc and E4APLuc reporter plasmids, P. R. Dobner for the AP-1-pCMV2 expression plasmids, S. Mcknight for the MSV/EBPβ-pBluescript plasmid, S. Akira for the pBlue610 plasmid, A. Nigh and M. E. Greenberg for FosB−/− mice, and P. R. Dobner and C. E. Schrader for helpful criticism of the manuscript.
↵1 This work was supported by National Institutes of Health Grant AI42108 (to J.S.) The contents of the paper are solely the responsibility of the authors and do not necessarily represent the official views of the National Institute of Allergy and Immunology.
↵2 Current address: Department of Biology, 68-265, Massachusetts Institute of Technology, Cambridge, MA 02139.
↵3 Address correspondence and reprint requests to Dr. Janet Stavnezer, Department of Molecular Genetics and Microbiology, University of Massachusetts Medical School, 55 Lake Avenue North, Worcester, MA 01655-0122. E-mail address:
↵4 Abbreviations used in this paper: GL, germline; CD40L, CD40 ligand; C/EBP, CAAT/enhancer-binding protein; HPRT, hypoxanthine phosphoribosyltransferase; Act D, actinomycin D; DRB, 5,6-dichlorobenzimidazole riboside; Oct, octamer-binding protein; BSAP, B cell-specific activator protein; MMLV, Moloney murine leukemia virus; wt, wild-type; mt, mutated, mutant.
- Received January 10, 2000.
- Accepted October 9, 2000.
- Copyright © 2001 by The American Association of Immunologists