GLT of IgG2b constant gene
Class switch recombination
- S region:
Transforming growth factor (TGF)-β1 directs class switch recombination (CSR) to IgG2b as well as to IgA. Smad3/4, Runx3 and p300 mediate TGF-β1-induced germ-line (GL) α transcription leading to IgA expression. However, the molecular mechanisms by which TGF-β1 induces IgG2b CSR are unknown. We used luciferase reporter plasmids to investigate how TGF-β1 regulates the activity of the promoter for GL transcripts of IgG2b constant gene (GLγ2b promoter). Similarly to the GLα promoter, overexpression of Smad3/4 and Runx3 enhances TGF-β1-induced GLγ2b promoter activity. Mutation analysis of the promoter identified likely Smad- and Runx3-binding sites. Also similar to the GLα promoter, overexpression of p300 enhances Smad3/4-mediated promoter activity, whereas E1A represses promoter activity. Since these regulation mechanisms underlying both GLα and GLγ2b transcription are similar, we explored the possibility that TGF-β1 induces IgA CSR via transitional IgG2b CSR. TGF-β1 enhances the expression of both Iα-Cμ and Iα-Cγ2b circle transcripts, indicative of direct (Sμ→Sα) and sequential CSR (Sμ→Sγ2b→Sα).
Ig class switch recombination (CSR) enables the recombined variable region gene segment (VDJ) to be expressed with a new downstream heavy chain constant region (CH) gene. Class switching is effected by a deletional recombination which occurs between switch region (S region) sequences located upstream of each of the CH genes except Cδ. CSR is directed to a particular CH gene by cytokines that induce transcription from germ-line (GL) CH genes before switch recombination to the same CH gene 1. CSR occurs via deletional recombination and excision of the intervening DNA between the two S regions as a switch circle 2.
It is now clear that CSR requires GL transcription through target S regions 1, 3 and expression of activation-induced cytidine deaminase (AID) 4. AID is proposed to either edit unknown mRNA to generate endonucleases or act directly on DNA, resulting in CSR 5. Based on the latter model, AID deaminates deoxycytidine residues in single-strand DNA to generate deoxyuridine residues 6, and it has been suggested that, in conjunction with uracil-DNA glycosylase and base excision repair 7, AID causes the cleavage of DNA leading to CSR 8. The data indicate that GL transcription is involved in creation of the single-strand DNA target for AID 9. In addition, the S region of the GL transcript (GLT) appears to form an R-loop structure, perhaps causing the non-template strand to become single-stranded and therefore a target for AID 10. These results are thought to explain, wholly or in part, why GL transcription is essential for CSR.
TGF-β1 induces both GLα transcription and subsequent CSR to IgA 11 and IgG2b 12. Several groups have analyzed the molecular mechanisms underlying GLα transcription induced by TGF-β1. Smad3/4 and Runx3 mediate TGF-β1-induced GLα promoter activity 13, 14. In addition, an Ets-binding site was shown to be essential for activity, and overexpressed Ets1 and Fli-1 enhance promoter activity synergistically with Runx 15, 16. In addition, p300 cooperates with Smad3/4 and Runx3 to activate the GLα promoter, whereas E1A inhibits these cooperative effects 17. Furthermore, TGF-β1-induced GLα transcription results in IgA secretion 18. In contrast, the molecular mechanisms underlying TGF-β1-induced GL transcription of IgG2b constant gene (GLγ2b transcription) are entirely unknown, although it is clear that the Iγ2b promoter and exon sequences are necessary for both GL gene transcription and subsequent CSR to IgG2b 19.
To elucidate molecular mechanisms by which TGF-β1 induces GLγ2b transcription, we constructed mouse GLγ2b promoter reporters and analyzed their activity. We report here that, similarly to the GLα promoter, Smad3/4 mediate TGF-β1-induced GLγ2b promoter activity and that Runx3 and p300 act as co-activators. Finally, due to the similarity between regulation of GLα and GLγ2b transcription, we asked whether TGF-β1-induced GLγ2b transcription and CSR are linked to IgA CSR. We found that IgA CSR can occur sequentially through IgG2b CSR, and that Smad3/4, Runx3 and p300 mediate this process.
Involvement of TGF-β1 and Smad proteins in GLγ2b transcription
To begin analysis of the promoter for GLγ2b transcription, we searched the nucleotide sequence of the segment containing the RNA initiation site(s) for potential transcription factor-binding elements 20. As shown in Fig. 1A, there are multiple putative Smad-binding elements (SBE), Runx-binding elements (RBE), and an Ets-binding element (EBE) within the tentative GLγ2b promoter region (TFSEARCH Version 1.3, Parallel Application TRC Laboratory, RWCP, Japan; and MatInspector Version 3.0, Genomatix Software). To examine whether this DNA segment did indeed have promoter activity, we constructed two luciferase reporter plasmids, pG2b.1 and pG2b.2, which differed by the amount of segment included 3′ to the RNA initiation site (Fig. 1A). We observed that TGF-β1 treatment increased the pG2b.1 promoter activity by twofold but not pG2b.2 activity (Fig. 1B). Since TGF-β activation of the GLα promoter is mediated by Smad3/4, we tested the effect of these proteins on pG2b.1 reporter activity (Fig. 1C). Overexpression of Smad3/4 further increased the TGF-β1-induced pG2b.1 promoter activity, suggesting that Smad3/4 mediate TGF-β1-induced GLγ2b transcription.
To assess whether Smad3/4 are involved in endogenous GLγ2b transcription, we transfected LPS-activated mouse spleen cells with Smad3/4, and examined the expression of endogenous GLγ2b transcripts. TGF-β1 treatment increased the GLγ2b transcription, and this effect was further augmented by overexpressed Smad3/4 (Fig. 1D). In contrast, transfected Smad3/4 diminished GLγ1 transcription in the presence of TGF-β1.
To obtain additional evidence that Smad3 is involved in GLγ2b transcription, we tested the effect of a dominant-negative (DN) form of Smad3 (DN-Smad3) on GLγ2b reporter activity. Overexpression of DN-Smad3 inhibited TGF-β1-induced GLγ2b promoter activity (Fig. 2A), indicating that Smad3 mediates TGF-β1-induced GLγ2b transcription. Two other reporters, GLα Luc and p3TP-Lux, which are also known to be stimulated by Smad3 signaling 14, 21, served as positive controls.
Both phosphorylated Smad2 and Smad3 are known to form heterodimeric complexes with Smad4. Overexpression of either Smad3/4 or Smad2/4 enhances TGF-β-induced p3TP-Lux promoter activity 21. Furthermore, Smad2, Smad3 and Smad4 synergize to activate the TGF-β-inducible plasminogen activator inhibitor-1 promoter 22. By contrast, others demonstrate antagonism between Smad2 and Smad3 on target genes 23, 24. Therefore, it was important to determine the effect of Smad2 on GLγ2b promoter activity. The effect of Smad2/4 on GLγ2b promoter activity was less potent than that of Smad3/4, and overexpression of Smad2 actually hindered the Smad3/4-mediated GLγ2b reporter activity (Fig. 2B). A similar effect of Smad2 on GLα promoter activity was observed, although it was not as striking, consistent with previous observations 18.
Characterization of putative SBE within GLγ2b promoter region
To determine which of the four potential SBE are important for promoter activity, we tested the effect of deletion of each of the CAGA elements in transient reporter assays in L10A6.2 cells. Mutant reporters (mS1, mS2 and mS4) responded to TGF-β1 and overexpressed Smad3/4 similarly or better than wild-type pG2b.1 (Fig. 3A). However, the reporter activities of mS3 and mS3.4 virtually disappeared under the same conditions. These results imply that the third CAGA is the essential SBE in the GLγ2b promoter. Therefore, we investigated binding of nuclear proteins from TGF-β1-treated L10A6.2 cells to a probe containing the third CAGA sequence (SBE3) by electrophoretic mobility-shift assay (EMSA) (Fig. 3B). TGF-β1 treatment enhanced the DNA-protein complex, which was eliminated by pre-incubation with a tenfold excess of the cold probe.
Based on these results, it was possible that the upstream GLγ2b segment containing the first and the second CAGA sequences might be dispensable for promoter activity. To test this, we examined the expression of the pG2b.3 reporter, which contains nucleotides –139 to +254 and lacks putative SBE1 and SBE2. The basal promoter activity of pG2b.3 is higher, whereas its inducibility is lower than that of pG2b.1 (Fig. 3C). However, the overall promoter activity of pG2b.3 is similar to that of pG2b.1, indicating that the DNA segment –434 to –140 is not required for optimal TGF-β-induced GLγ2b promoter activity.
Roles of Runx3 and p300 for GLγ2b promoter activity
In addition to the putative SBE, there are putative RBE and EBE within the GLγ2b promoter (Fig. 1A). As Runx3 is important for GLα transcription, we examined whether Runx3 plays a role in TGF-β1-induced GLγ2b promoter activity. Unlike the GLα promoter 13, overexpressed Runx3 alone had no affect on either basal or TGF-β1-induced promoter activity (Fig. 4A). However, overexpression of Runx3 synergistically augmented the effect of overexpressed Smad3/4. To attempt to experimentally identify RBE within the GLγ2b promoter, we constructed four different mutant reporters from which various putative RBE sites were deleted, as diagrammed in Fig. 4B. Mutation of putative RBE2 eliminated all promoter activity (Fig. 4B), although mutation of the three other putative RBE had no effect. These results indicate that RBE2 (TGTGGGT, +41 to +47) is essential for GLγ2b promoter activity.
An Ets-binding site is essential for GLα promoter activity 16; so we were interested to determine whether the potential EBE at nucleotides –4 to +2 in the GLγ2b promoter is important for expression. However, neither deletion of this element from pG2b.1 nor overexpression of Elf-1 (one of the Ets family members) had any effect on promoter activity or response to TGF-β1 or to Smad3/4 (data not shown).
p300 interacts with Smad3 to enhance transactivation by Smad 25 and acts as a co-activator in GLα promoter activity 17. Therein, we assessed the effect of p300 on GLγ2b promoter activity (Fig. 5). Overexpressed p300 enhanced basal and TGF-β1-induced promoter activity, and the activity was further enhanced by cotransfection with Smad3/4, while overexpression of E1A, an adenoviral oncoprotein that inhibits action of p300, abolished the combined effect of Smad3/4 and p300 by 70%. Thus, similar to its activity on GLα transcription, p300 acts as a co-activator, and E1A as a repressor of GLγ2b transcription mediated by Smad3/4.
Implication of TGF-β1-induced IgG2b class switching for IgA expression
Thus far, our data reveal that Smad3/4, Runx3 and p300 synergistically augment TGF-β1-induced GLγ2b transcription. These same proteins mediate TGF-β1-induced endogenous GLα transcription, which in turn leads to IgA expression 17. As it has been shown that IgA isotype switching occurs both directly and sequentially 26, 27, we asked whether TGF-β1 induces IgA CSR via intermediate IgG2b CSR.
After CSR to IgA, the GLμ promoter becomes associated with the Cα gene and continues to be active, generating transcripts termed post-switch transcripts (Iμ-Cα or PSTIμ-Cα) 4, 28 (Fig. 6A). Furthermore, the DNA sequences between Sμ and Sα are looped out of the chromosome as switch circles during CSR, and another type of transcript, termed circle transcript (CT), in this case consisting of the Iα exon spliced to the Cμ exon (Iα-Cμ or CTIα-Cμ), is transcribed from the switch circle due to the active Iα promoter 29 (Fig. 6A). Thus, one may monitor CT expression as indicative of active CSR. If sequential switching from IgG2b to IgA occurs, CTIα-Cγ2b could be generated as shown in Fig. 6A. Therefore, CTIα-Cγ2b can serve as a molecular marker of sequential switching to IgA via IgG2b, while CTIα-Cμ can serve as a marker of direct switching to IgA.
We first tested the effect of TGF-β1 on the expression of GLT, PST and CT by LPS-stimulated spleen cells (Fig. 6B). TGF-β1 increased the expression of GLα (GLTIα-Cα) and GLγ2b (GLTIγ2b-Cγ2b) but not GLγ3 transcripts (GLTIγ3-Cγ3). Similarly, TGF-β1 increased the expression of PSTIμ-Cα and PSTIμ-Cγ2b but not PSTIμ-Cγ3. Furthermore, TGF-β1 enhanced the expression of both CTIα-Cμ and CTIα-Cγ2b, but TGF-β1 decreased CTIα-Cγ3 expression. These results indicate that switching to IgA induced by TGF-β1 is both sequential (Sμ→Sγ2b→Sα) and direct CSR (Sμ→Sα).
Finally, we examined the effect of Smad3/4, Runx3 and p300 on the expression of GLT and CT (Fig. 6C). Transfection of Smad3/4, Runx3 and p300 further increased the expression of GLTIα-Cα and GLTIγ2b-Cγ2b, and also the expression of CTIα-Cμ and CTIα-Cγ2b. We confirmed the junctional sequences of PCR products corresponding to each GLT, PST and CT from the experiment shown in Fig. 6 (Fig. 7). For example, junctional sequences of CTIα-Cγ2b agree to the reported ones (GenBank Accession No.: Iα exon, D11468; Cγ2b exon, D78344). These results demonstrate that TGF-β1 directs CSR to IgA both directly and sequentially through transitional IgG2b CSR, and that the effect of TGF-β1 is mediated by Smad3/4, Runx3 and p300.
The findings herein extend earlier studies on the effect of TGF-β1 on IgG2b isotype switching, in which it was shown that TGF-β1 directs CSR to IgG2b, in addition to IgA, in splenic B cells 12. Thus, the present study demonstrates that Smad3, Smad4 and Runx3 are important mediators in TGF-β1-induced GLγ2b transcription leading to IgG2b isotype switching. Although there are several potential SBE and RBE within the GLγ2b promoter region, only one SBE (CAGA, –38 to –35) and one RBE (TGTGGGT, +41 to +47) were shown to be critical for GLγ2b promoter activity. This result differs from the results for the GLα promoter, that two tandem combined Smad-Runx elements are required for induction by TGF-β1 14, 16, 30, 31.
Overexpression of Smad3/4 increases GLγ2b promoter activity. In contrast, overexpression of Runx3 alone did not affect promoter activity, although it further augmented promoter activity when cotransfected with Smad3/4. These results reveal that the action of Runx3 is dependent on the existence of Smad3/4, indicating that Runx3 is a co-activator for Smad3/4 in GLγ2b transcription. However, the RBE +41 to +47 (TGTGGGT) is indispensable for overall promoter activity, as both basal and Smad3/4-mediated promoter activities were dramatically diminished when this element was deleted. The poor activity of the pG2b.2 reporter, which contains the GLγ2b segment –434 to +10, is likely to be at least partially due to the deletion of this essential RBE.
We note that responsiveness of pG2b.1 to TGF-β1 is lower than that of either GLα Luc or p3TP-Lux. This result suggests that expression of GLγ2b transcript does not require added TGF-β1, unlike GLα transcription, although Smad3/4 clearly further stimulates TGF-β1-induced GLγ2b transcription. In this context, it is noteworthy that LPS alone readily induces the expression of endogenous GLTIγ2b-Cγ2b, as previously reported 32, and PSTIμ-Cγ2b but very little GLTIα-Cα or PSTIμ-Cα (Figs. 1, 6). These results are consistent with the fact that IgG2b CSR tends to take place in activated B cells without exogenous cytokine addition, perhaps due to low levels of endogenous TGF-β1 12.
There are differences between the GLα and GLγ2b promoters that might make the GLα promoter more dependent upon and responsive to TGF-β1. One example is the finding that overexpression of Runx factors increases TGF-β1 induction of the GLα promoter but not the γ2b promoter (Fig. 4) 13, 15. Furthermore, the transcription factor CREB/ATF synergizes with Smad3/4/Runx3 to promote GLα transcription, and mutation of the GLα CREB/ATF-binding site has been shown to reduce responsiveness to TGF-β1 13, 14, 16, 30, 31. Interestingly, a recent study has demonstrated that after activation by TGF-β, the Smad3/4 heterodimer directly activates protein kinase A, which in turn stimulates DNA binding by CREB 33. It is unknown whether CREB is involved in the regulation of GLγ2b transcription.
Although GLα and GLγ2b transcription are not identically regulated, there appear to be many similarities. The functions of Smad3/4, Runx3, p300 and E1A, as well as Smad2, are quite comparable for activities of the GLγ2b and GLα promoters. Consistent with these results, we found that TGF-β1 increases the expression of both CTIα-Cμ and CTIα-Cγ2b, indicating that TGF-β1 induces IgA isotype switching through direct switching (IgM→IgA), as well as a sequential switching (IgM→IgG2b→IgA). Interestingly, TGF-β1 does not stimulate sequential switching from all IgG subclasses, as expression of CTIα-Cγ3, a marker for sequential switching from IgG3→IgA, is down-regulated by TGF-β1. Furthermore, our data indicate that recombination between Sγ2b and Sα regions does not occur prior to IgG2b expression, because CTIγ2b-Cα was not induced by TGF-β1 (data not shown). Therefore, TGF-β1-induced sequential switching to IgA is attributed to successive Sμ→Sγ2b→Sα recombination, i.e. IgM→IgG2b→IgA.
Studies by other investigators have previously shown that IgA and IgE expression is mediated by sequential switch recombination. TGF-β was shown to induce IgA CSR through both direct Sμ→Sα and sequential Sμ→Sγ→Sα DNA recombination in mouse and human B cells 26, 27. Furthermore, IL-4 directs CSR to IgE predominantly through successive Sμ→Sγ1→Sϵ recombination in mouse 34–36 and Sμ→Sγ4→Sϵ in human 37–39. Consistent with our results, the GLγ1 and GLϵ promoters have been shown to be similarly but not identically regulated, with the GLϵ promoter and IgE CSR more dependent on IL-4 than the γ1 promoter and IgG1 CSR 40, 41. However, it still remains to be elucidated why it is evolutionarily advantageous for a single cytokine (TGF-β1) to induce direct switching to both IgA and IgG2b isotypes, which appear to be so different in function. Perhaps these isotypes have unrecognized functional similarities, and it is also possible that the transitional induction of IgG2b potentially increases the level of IgA expression.
Materials and methods
BALB/c mice were purchased from B&K Universal Co. (Fremont, CA) and were maintained on Purina Laboratory Rodent Chow 5001 (Ralston Purina Co., Richmond, IN) ad libitum, in an animal environmental control chamber (Myung-Jin Inst. Co., Seoul, Korea). Eight- to twelve-week-old mice were used, and animal care was in accordance with the institutional guidelines of Kangwon National University.
Genes encoding Smad2 42, Smad3 43 and Smad4 44, subcloned into Flag-pcDNA3 45, were provided by Dr. M. Kawabata (The Cancer Institute, Tokyo, Japan). The DN-Smad3 expression plasmid (Smad3D407E) 21 was provided by Dr. M. Kato (The Cancer Institute). The expression plasmid for Runx3 46 was obtained from Dr. S. Hiebert (St. Jude Children's Research Hospital, Memphis, TN). The expression plasmid for p300 47 was provided by Dr. R. Janknecht (Mayo Graduate School, Rochester, MN). E1A 48, subcloned into pcDNA3, was provided by Dr. M. Kawabata (The Cancer Institute). Mouse cDNA for Elf-1 cloned in pCDNA3 49 was obtained from Dr. M. Roussel (St. Jude Children's Research Hospital).
Reporter plasmid constructs
Three different GLγ2b promoter DNA fragments (–434 to +254, –434 to +10 and –139 to +254) were amplified from mouse spleen genomic DNA by PCR. The PCR primers were derived from the GLγ2b promoter nucleotide sequences previously reported 20, 50 (Accession No. M19414 and D78344). The GLγ2b promoter segments were subcloned into pGL3 (Promega, Madison, WI). GLγ2b promoter reporters for –434 to +254, –434 to +10 and –139 to +254 were named pG2b.1, pG2b.2 and pG2b.3, respectively. Mutations were introduced into pG2b.1 using QuikChangeTM Site-Directed Mutagenesis (Stratagene, La Jolla, CA). Reporters containing mutations of the putative SBE and putative RBE were constructed. The GLα –130 to +14 Luc and GLγ1 –148 to +202 Luc promoters were subcloned into pXP2 13, 51. p3TP-Lux 52 was obtained from Dr. J. Massagué (Memorial Sloan-Kettering Cancer Center, New York, NY).
Transfection and luciferase assays
Transfection was performed by electroporation with a Gene Pulser II (Bio-Rad, Hercules, CA) as described 18. Reporter plasmids were cotransfected with expression plasmids and pCMVβgal (Stratagene), and luciferase and β-gal assays performed as described 18.
RNA preparation, reverse transcription, and PCR were performed as described previously 18. Primers for PCR were synthesized by Bioneer Corp. (Seoul, Korea), i.e. GLTIα-Cα sense 5′-CTACCATAGGGAAGATAGCCT-3′ and antisense 5′-TAATCGTGAATCAGGCAG-3′ (product size 206 bp); GLTIγ2b-Cγ2b sense 5′-GGGAGAGCACTGGGCCTT-3′ and antisense 5′-AGTCACTGACTCAGGGAA-3′ (product size 318 bp); GLTIγ3-Cγ3 sense 5′-CAAGTGGATCTGAACACA-3′ and antisense 5′-GGCTCCATAGTTCCATT-3′ (product size 349 bp); GLTIγ1-Cγ1 sense, 5′-CAGCCTGGTGTCAACTAG-3′ and antisense 5′-CTGTACATATGCAAGGCT-3′ (product size 532 bp); PSTIμ-Cα sense 5′-ACCTGGGAATGTATGGTTGTGGCTT-3′ and antisense 5′-TAATCGTGAATCAGGCAG-3′ (product size 224 bp); PSTIμ-Cγ2b sense 5′-ACCTGGGAATGTATGGTTGTGGCTT-3′ and antisense 5′-AGTCACTGACTCAGGGAA-3′ (product size 246 bp); PSTIμ-Cγ3 sense 5′-ACCTGGGAATGTATGGTTGTGGCTT-3′ and antisense 5′-GGCTCCATAGTTCCATT-3′ (product size 297 bp); CTIα-Cμ sense 5′-CTACCATAGGGAAGATAGCCT-3′ and antisense 5′-TCTGAACCTTCAAGGATGCTCTTG-3′ (product size 365 bp); CTIα-Cγ2b sense 5′-CCTAAGCTCTCTACCATAGG-3′ and antisense 5′-GTCACGGAGGAACCAGTTGT-3′ (product size 195 bp); CTIα-Cγ3 sense 5′-CTACCATAGGGAAGATAGCCT-3′ and antisense 5′-GGCTCCATAGTTCCATT-3′ (product size 249 bp); β-actin sense 5′-CATGTTTGAGACCTTCAACACCCC-3′ and antisense 5′-GCCATCTCCTGCTCGAAGTCTAG-3′ (product size 320 bp).
All reagents for RT-PCR were purchased from Promega. PCR reactions for β-actin were performed in parallel in order to normalize cDNA concentrations within each set of samples. Aliquots of the PCR products were resolved by electrophoresis on 2% agarose gels.
PCR products for GLT, PST and CT were purified by using GENECLEAN® Turbo kit (Qbiogene, Carlsbad, CA), and their sequences were determined by Macrogen (Seoul, Korea). The primers for the sequencing analysis were as described above.
Preparation of oligonucleotide probes and nuclear extracts for EMSA
The double-stranded (ds) oligonucleotide probe for EMSA was prepared as described 16. The upper strand sequence for the GLγ2b ds oligonucleotide probe (–42 to –15) was 5′-ACCCCAGACACTGAAACCAACAGAAGAA-3′. 32P-end-labeled oligonucleotides were prepared using the Gel Shift Assay Systems (Promega). Nuclear extracts from L10A6.2 cells were prepared as described 16.
DNA binding reactions were performed in 20-μl reaction volumes containing 50 fmol of end-labeled ds DNA probe and 2 μg of nuclear extract in buffer of Gel Shift Assay Systems (Promega). The reaction mixtures were incubated at room temperature for 30 min and then loaded onto 6% native polyacrylamide gels and electrophoresed in 0.5× TBE buffer at 130 V for 4 h. For competition experiments, cold probe (a tenfold molar excess of unlabeled oligonucleotides over the labeled probe) was added to the entire mixture, except the probe that was added last.
This work was supported by a Vascular System Research Center grant from the Korea Science & Engineering Foundation to P.-H.K. and by a grant from the NIH (AI23283) to J.S.