CBFα3 (AML2) Is Induced by TGF-β1 to Bind and Activate the Mouse Germline Ig α Promoter¹

Upon antigenic or mitogenic stimulation, B lymphocytes proliferate and differentiate toward Ab secretion. As they proliferate, some of the progeny switch from expression of IgM to expression of other Ig classes. Ig class switching allows the recombined variable (VDJ) region gene of the heavy chain to be expressed with a new downstream constant region (C_H)³ gene, resulting in a change in the effector function of the Ab. Class switching is effected by a deletional recombination that occurs between switch regions located upstream of each C_H gene. Class switch recombination is directed to particular C_H genes by cytokines that induce transcription of germline (GL) transcripts from unrearranged C_H genes before switch recombination (1). Although the functions of GL transcripts are unknown, transcription of unrearranged C_H genes is important for regulation of class switch recombination. Two types of evidence support this statement. First, there is a good correlation between the specificity of GL transcription and the specificity of isotype switching events. Second, deletion of a DNA segment containing the promoter and first exon (I or GL exon) of the GL transcript by knockout technology abrogates switching to that C_H gene (2, 3, 4, 5).

TGF-β1 induces GL α transcripts and directs class switching to IgA in mouse splenic B cells activated by polyclonal activators, such as LPS (6, 7). Similarly, TGF-β1 increases GL α transcripts, and in the presence of LPS induces switching to IgA in the surface IgM⁺ (sIgM⁺) I.29μ B lymphoma cell line (8, 9). The induction of GL α transcripts by TGF-β1 in I.29μ cells is due to regulation at the transcriptional level (8). Previous work showed that the promoter segment containing nucleotides from −128 to +46, relative to the first initiation site for GL α transcripts, is sufficient for expression and TGF-β1 inducibility of a reporter gene in two different B cell lines (10). Within the −128/+46 segment, a novel element (called TβRE) was shown to be required for TGF-β1 inducibility of a reporter gene and, when multimerized, to be sufficient to transfer the inducibility to a minimal c-fos promoter (10). The TβRE is also present in the human GL α1 and α2 promoters and has been shown to be responsive to TGF-β1 in the human GL α1 promoter (11, 12). Despite the importance of the TβRE for GL α transcription, no protein that binds the TβRE and regulates the GL α promoter has been identified.

The core-binding factor (CBF) family of transcription factors plays a key role in hematopoiesis and bone ossification. The family consists of three α (CBFα) and one β (CBFβ) subunits (reviewed in Refs. 13 and 14). CBF binds to DNA as a heterodimer of the α and β subunits. All the α subunits contain a highly conserved runt domain that is responsible for both DNA binding and dimerization with the CBFβ subunit (15, 16, 17, 18, 19, 20), which by itself cannot bind DNA but increases the affinity of CBFα protein for DNA about fivefold (16, 17, 18). CBFα is also known as acute myeloid leukemia (AML) or polyomavirus enhancer core-binding protein-2α (PEBP2α). Thus, CBFα1 is also known as AML3 or PEBP2αA, CBFα2 as AML1 or PEBP2αB, and CBFα3 as AML2 or PEBP2αC. CBFβ is also known as PEBP2β.

CBF specifically recognizes the consensus sequence ACCPuCA (13). This sequence motif has been identified in both viral and cellular promoters and/or enhancers of a number of genes including polyomavirus and murine leukemia virus and some hematopoietic-specific genes, such as in the Ig μ intron enhancer (21), and the genes for the TCR (TCR α-, β-, δ-, and γ-chains), CD3ε, cytokines (IL-3 and granulocyte-macrophage CSF), the CSF receptor, and the myeloid-specific genes myeloperoxidase and neutrophil elastase (reviewed in 13 . The binding site has also been found in bone-specific genes (22, 23).

Although all the α subunits bind and activate transcription via the same sequence motif (16, 19, 24, 25, 26, 27), the expression levels of different CBFα proteins differ among different tissues. CBFα1 (AML3) is most abundant in osteoblasts and is also present in thymocytes and T cell lines (16, 27, 28, 29). CBFα2 (AML1) is abundant in T and myeloid cells and also in B cells and some B cell lines (15, 26, 28, 30, 31). CBFα3 (AML2) is most abundant in B and myeloid cell lines (26). CBFβ is ubiquitously expressed (17, 18, 28). Thus, at least one explanation for the different functions of the three CBFα proteins in vivo must be their different patterns of expression. A knockout of the CBFα1 gene in the mouse completely blocks bone ossification. Such mice die just after birth due to inability to breathe (32). A knockout of the CBFα2 gene in the mouse results in death on days 12–14 of fetal development owing to hemorrhaging in the central nervous system and blockage of fetal liver hematopoiesis (33, 34). The targeted deletion experiments indicate that the function of CBFβ is similar to that of CBFα2, as it is also essential for definitive hematopoiesis of all lineages and its deletion leads to death at midgestation (35, 36, 37). Furthermore, the CBFα2/CBFβ transcription factor complex represents one of the most important targets of chromosomal translocations found in human leukemias, hence the alternative name AML (13, 14, 38).

In this report, we demonstrate that CBF (AML) proteins bind to the TβRE present within the GL α promoter for GL α transcripts and that CBFα3 (AML2) and TβRE binding activities are up-regulated by TGF-β1. We also demonstrate that CBF activates the GL α promoter.

Three mouse B lymphoma cell lines were used in this study: I.29μ (22D subclone) and CH12.LX, two sIgM⁺ B cell lines that can be induced to undergo class switch recombination to IgA (39, 40, 41, 42), and A20.3, a sIgG2a⁺ B cell line that has not been demonstrated to undergo class switching (43, 44).

A20.3 and CH12.LX cells were cultured at 37°C in an atmosphere of 5% CO₂ in RPMI 1640 medium supplemented with 10% FBS, 50 μM 2-ME, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 0.1 mg/ml kanamycin sulfate (the last six reagents were from Life Technologies Laboratories, Grand Island, NY). I.29μ cells were cultured similarly to A20.3 cells, except in 8% CO₂, and the medium was supplemented with 0.1 U/ml regular purified pork insulin (Squibb and Sons, Princeton, NJ), and the FBS concentration was 20%. In transfection experiments, human TGF-β1 (R & D System, Minneapolis, MN) was added right after transfection at 1–2 ng/ml. The amount of TGF-β1 used was determined to be in the optimal range (data not shown).

Splenic B cells were purified by depletion of T cells, using mAbs for Ty1.2 (HO13.4), Thy1 (Jij.10), CD4 (GK1.5), and CD8 (3.168) followed by addition of anti-rat κ-chain mAb (MAR 18.5) and guinea pig complement. Dead cells were removed by centrifugation on Ficoll-Hypaque gradients. Purified splenic B cells were cultured as were A20.3 cells, with addition of LPS (50 μg/ml) for 18 h and TGF-β1 (2 ng/ml) for 14 h.

The sequences of the top strand of the double-stranded oligonucleotides used in EMSAs are given below. Lower case letters indicate nucleotides not present in the α promoter that were added for cloning. Underlines indicate consensus motifs for the transcription factor binding sites. Bold nucleotides indicate mutations: CBF consensus, CGTATTAACCACAATACTCG; TβRE(−130/−104), CACCACAGCCAGACCACAGGCCAGACA; m₁-TβRE, gaTCCACCACAGAACTACCACAGGAACTACA; m₂-TβRE, gaTCCGTGACAGACAGACCTGCAGCCAGACA; Cα(−104/−69), ATGACGTGGAGGCAAGCGGCCACAACGTGGAGGTGG; Cα(−94/−55), GGCAAGCGGCCACAACGTGGAGGTGGAACAGGAAGTGGGT; Cα(−95/−78), AGGCAAGCGGCCACAACG; Ets(Cα −71/−55), TGGAACAGGAAGTGGGT; and reverse PCR primer (+14/−7), cgcggatccTGGATGGGTCAGACTGAGCGC. Single-stranded oligonucleotides were obtained from Operon Technologies (Alameda, CA).

Wild-type or mutated GL α promoter −130/+14 segments were created by PCR using oligonucleotides containing wild-type or mutated TβREs as primers. Some of the segments with mutations were obtained using an overlap extension technique (45). PCR-amplified fragments were then cloned into SmaI/BglII sites upstream of the luciferase reporter gene in the pXP2 vector (46). Each construct was sequenced to verify that only the desired mutations had been introduced. DNA sequencing was performed using the dideoxy chain termination method with Sequenase version 2.0 kits (U.S. Biochemical Corp. Cleveland, OH).

Expression plasmids containing cDNAs for mouse CBFα1 (AML3), human CBFα2 (AML1B), the trans-activating form of AML1 (25), and human CBFα3 (AML2) (26) were obtained from S. Hiebert, St. Jude Children’s Research Hospital (Memphis, TN). Expression plasmids containing cDNAs for mouse CBFα2 and CBFβ were obtained from N. Speck, Dartmouth Medical School (Hanover, NH). These expression plasmids (30 μg) were cotransfected with the luciferase reporter plasmids driven by the mouse GL α promoter −130/+14. At the amino acid level, human and mouse CBFα3 are identical (47), whereas human and mouse CBFα2 share about 90% identity (25, 48). We compared the activities of human and mouse CBFα2 expression plasmids and found no difference (data not shown). Since the CBF cDNAs were cloned into different expression vectors, we first tested whether the different empty expression vectors have different effects on the α promoter activity. We cotransfected the empty expression vectors pCMV-2, pCMV-5, and pcDNA I into A20.3 cells and found that all have similar luciferase activities. Therefore, we used pCMV-2 as the vector control for all CBF expression plasmids.

For each transfection, 50 × 10⁶ cells were resuspended in 1 ml of RPMI 1640 containing 50 μg of luciferase reporter plasmid DNA for I.29μ and 30 μg for A20.3 cells along with 10 μg of control plasmid DNA before imposing an electric shock at 1250 μF and 750 V/cm using Cell ZapII (Anderson Electronics, Brookline, MA). To control for variation in transfection efficiency, the plasmid pSV2CAT, which expresses the CAT gene under control of the SV40 enhancer/promoter, was cotransfected in all experiments in I.29μ cells, and the plasmid pPGKβ-gal, which expresses the β-gal gene under control of the PGK enhancer/promoter, was used in all experiments in A20.3 cells. After transfection, cells were rested at room temperature for 10 min and then pipetted into complete FBS/RPMI medium at approximately 1.7 × 10⁶ cells/ml. Cells were incubated for 24 h at 37°C and then assayed for luciferase, CAT (49), and β-gal (50) activities. CAT activity was measured by a liquid scintillation counter after 20-h incubation at 37°C.

Double-stranded (ds) oligonucleotides were prepared and labeled as previously described (51).

Cells were pelleted and resuspended in fresh medium at 10⁶ cells/ml before adding stimuli. After incubation, cells were harvested, and cytoplasmic and nuclear extracts were prepared using a modification of the small scale method of Schreiber et al. (49, 52). Protein concentrations of the cytoplasmic and nuclear extracts were determined using the Bradford assay (Bio-Rad, Richmond, CA).

DNA binding reactions were performed in 15-μl reaction volumes containing 0.1–1 ng (15,000–30,000 cpm) of end-labeled dsDNA probe, 1–5 μg of nuclear extracts, and 2–4 μg of poly(dI-dC) (Pharmacia, Piscataway, NJ). The final concentration of NaCl in each reaction mixture was adjusted to 80 mM by adding buffer C (52). The reaction mixtures were incubated at room temperature for 30 min and then loaded onto 4–6% native polyacrylamide gels. All gels were electrophoresed in recirculating 0.5× TBE buffer at 100–150 V for 2–4 h. For supershift experiments, Ab (or antiserum; 1 μl) and nuclear extracts were added to the entire mixture before the probe. Polyclonal Abs against human CBFα1 (AML3), CBFα2 (AML1B), CBFα3 (AML2), and CBFβ used in both EMSAs and Western blot analyses were obtained from S. Hiebert, St. Jude Children’s Research Hospital. mAb against mouse CBFβ from N. Speck, Dartmouth Medical School, was also used in Western blot analyses.

Western blot analyses were performed by standard procedures. Briefly, nuclear extracts (10 μg/lane) were resolved by 12% SDS-PAGE and electroblotted onto Millipore Immobilon-P transfer membrane. Membranes were incubated overnight at 4°C in PBS containing 5% milk and 0.1% Tween-20 and Abs. After a brief wash with the same solution without Ab, the membrane was incubated with the secondary Ab for 2–3 h at room temperature, followed by chemiluminescent detection using SuperSignal ULTRA Chemiluminescent Substrate (Pierce, Rockford, IL) according to the manufacturer’s specifications. Membranes were exposed for 10 s to 5 min to Kodak XAR film (Eastman Kodak, Rochester, NY) for detection of signals.

Quantitation of DNA binding activity and protein expression were performed by densitometry using a Molecular Dynamics Personal Densitometry SI (Sunnyvale, CA), and the results were analyzed using Bio-Rad MultiAnalyst.

Previous studies identified a novel TβRE at –128/−105 relative to the first RNA initiation site in the mouse GL α promoter that is required for induction of the GL α promoter by TGF-β1. When the TβRE was used as a probe in EMSAs, complexes were formed using nuclear extracts from I.29μ cells (10). To examine whether complexes binding the TβRE could be induced by TGF-β1, we performed EMSAs using the TβRE (−130/−104) as a probe and nuclear extracts from I.29μ cells treated with TGF-β1 for varying times. As shown in Fig. 1 A, two major complexes were formed with the TβRE, one of which was induced fourfold by TGF-β1 treatment for 12 h, but not for 6 h. Induction of the complex slightly decreased after treatment for 18 h.

Since results from the kinetics experiment demonstrate a slow induction of binding activity by TGF-β1, we reasoned that induction of the TβRE binding complex may require de novo protein synthesis. To test this, we performed EMSAs using nuclear extracts from I.29μ cells treated with TGF-β1 in the presence or the absence of the protein synthesis inhibitor cycloheximide. As expected, treatment with cycloheximide completely prevented the induction of complexes binding to the TβRE (Fig. 1 B). These data indicate that TGF-β1 induces a complex in I.29μ extracts that binds the Ig α TβRE of the GL α promoter and that induction of the complex requires de novo protein synthesis.

The α TβRE contains two copies of a sequence (underlined in Fig. 2), and deletion of one of these two repeats reduces the TGF-β1 inducibility by 50% (10). To identify the binding site(s) in the TβRE for the inducible complex, we made two different sets of mutations in the direct repeats of the TβRE (m1 and m2) and tested the effects of these mutations in EMSAs. Figure 2,A demonstrates that binding of the TGF-β1-inducible complex in I.29μ nuclear extracts to the wild-type TβRE probe is competed well by a 100-fold excess of unlabeled wild-type TβRE and competed nearly as well by a 100-fold excess of an oligonucleotide with the m1 mutation, but not by an oligonucleotide with the m2-TβRE mutation. These results indicate that the binding of the TGF-β1-inducible complex requires nucleotides that are mutated in the m2-TβRE mutant, the wild-type sequence of which matches the consensus CBF binding site ACCACA. This was confirmed by EMSAs using an oligonucleotide containing a consensus CBF binding site as a competitor (Fig. 2,A). Since CBF proteins preferentially bind a site containing the nucleotide A just upstream of the core motif ACCACA (13), the CBF oligonucleotide competes better than the TβRE (Fig. 2,A). These data demonstrate that the TGF-β1-inducible complex binds to the two CBF binding sites within the TβRE (Fig. 2).

Results from the competition experiment indicate that the CBF sites (mutated in m2) in the TβRE are essential for binding of the TGF-β1-inducible complex, whereas the nucleotides mutated in the m1 mutation might not be required. To further examine the importance of the nucleotides mutated in mutant m1, we used the m1-TβRE oligonucleotide as a probe in EMSAs incubated with either nuclear extracts from I.29μ cells (Fig. 2,B) or CBF recombinant protein (Fig. 2,C). The results show that the m1 mutation greatly reduces binding of the TGF-β1-inducible complex compared with that formed with the wild-type probe (Fig. 2,B). No corresponding complex was formed with the m2-TβRE probe (data not shown). By contrast, recombinant CBFα2 runt domain (which is conserved among the CBFα subunits) binds equally well to the wild-type and m1-TβRE probes (Fig. 2 C). The equivalent binding to the wild-type or m1 probe is not surprising, since CBFα proteins do not preferentially bind to a site with a G (as in the wild-type TβRE) instead of a T (as in m1) upstream of the ACCACA consensus motif (13). These data suggest that the nucleotides mutated in the m1 probe interact with another protein(s) that increases the binding of CBF to the TβRE in the wild-type promoter.

To identify the specific CBF proteins binding to the GL α TβRE, we tested Abs against all known members of the family in supershift EMSAs. As shown in Fig. 3, Ab against CBFα3 (AML2) partially supershifted the major complex. Abs against CBFα1 (AML3), CBFα2 (AML1), and CBFβ weakly supershifted the complex. While the Ab against CBFα2 weakly cross-reacted with α1 and α3 on Western blots, the Abs against CBFα1 and α3 were specific (26). These data suggest that CBFα3 (AML2) is the predominant α subunit in I.29μ B cells treated by TGF-β1, but that the other CBF α subunits and CBFβ may also be present and bind the TβRE.

To directly examine the expression and regulation of CBF in I.29μ cells, we performed Western analyses using cytoplasmic and nuclear extracts from cells untreated or treated with TGF-β1 for 14 h. CBFα1 was not detected by Western analyses (data not shown), consistent with previous reports that CBFα1 (AML3) is not in pre-B and sIg⁺ B cell lines (21, 26), although it is present in a plasmacytoma line (21). The weak supershift observed using anti-CBFα1 Ab in EMSAs may be due to the greater sensitivity of the EMSA.

As shown in Fig. 4 A, two isoforms of CBFα2 are detected in cytoplasmic extracts, but only the smaller form is detected in nuclear extracts. The size of CBFα2 in I.29μ nuclear extracts agrees with the 48.6-kDa form identified in a mouse B lymphoma cell line by Bae et al. (30). Treatment with TGF-β1 had almost no effect on the level of CBFα2 in the cytoplasm but very slightly reduced its expression in nuclear extracts. The detection of CBFα2 cannot be due to cross-reaction with CBFα1, since CBFα1 was undetectable, nor was it due to cross-reaction with CBFα3, since CBFα3 was highly inducible by TGF-β1 (see below).

Interestingly, TGF-β1 induced expression by about 18-fold of two isoforms of CBFα3 in nuclei (Fig. 4,B). Expression of CBFα3 in the cytoplasm was barely detectable. The approximately 50-kDa forms were the expected size for CBFα3 (20, 31, 47). CBFβ was constitutively expressed in both cytoplasm and nuclei of I.29μ cells, and the 28-kDa form was reproducibly induced by TGF-β1 in I.29μ cells (Figs. 4 C and 5B).

Thus, the induction of the TβRE-binding activity observed in EMSAs is consistent with the induction of CBFα3 (AML2) by TGF-β1 as detected by Western analysis. The induction of binding activity is low, however, relative to the large induction of CBFα3 protein. This may be due to constitutive expression of CBFα2, as it binds the identical sequence element and has a similar m.w.

I.29μ cells constitutively express GL α transcripts, which can be further induced by treatment with TGF-β1. We asked whether the ability of other B cells and B cell lines to express GL α transcripts correlates with the expression of CBF and its induction by TGF-β1. The expression and regulation of CBFα3 and CBFβ were examined in splenic B cells and in two additional B lymphoma cell lines: the sIgM⁺ line, CH12.LX, and the sIgG2a⁺ line, A20.3. CH12.LX cells constitutively express GL α transcripts, which can be further increased by treatment with TGF-β1, and the cells can be induced to undergo class switch recombination to IgA (41, 42). A20.3 cells do not express GL α transcripts and have not been found to undergo class switching (10) (our laboratory observations).

Western blot analyses of nuclear extracts from these B cells indicated that the constitutive expression of CBFα3 was highest in CH12.LX cells, lower in I.29 μ cells, and barely detectable in A20.3 or splenic B cells (Fig. 5,A). TGF-β1 treatment induced expression of CBFα3 in splenic B and in I.29 μ cells and marginally induced CBFα3 (1.7-fold) in CH12.LX cells, but not in A20.3 cells. Thus, CBFα3 can be induced in two B cell lines and in a B cell population that can be induced to express GL α transcripts and to switch to IgA, but not in a B cell line that cannot be induced by TGF-β1 to express GL α transcripts. In contrast with the expression of CBFα3, the levels of CBFβ in nuclear extracts from different B cells were similar (Fig. 5 B).

Several isoforms of CBFα3 are detected in these various B cells. CBFα mRNA always occurs in multiple, alternatively spliced forms, the sizes of which may be cell type-specific (15, 17, 18, 26, 28, 30, 31).

The TβRE binding activities of nuclear extracts from the various cells were examined by EMSA. As shown in Fig. 5 C, the activity was highest in CH12.LX and I.29μ cells, lower in splenic B cells, and lowest in A20.3 cells. Furthermore, binding activity was not induced by TGF-β1 in A20.3 and splenic B cells, but was induced in both I.29μ (2.5-fold) and CH12.LX (1.3-fold) cells. Although induction by TGF-β1 was lower in CH12.LX than in I.29μ cells, there was a greater constitutive binding activity in CH12.LX extracts. Lack of induction of TβRE binding activity in splenic B cells may be due to constitutive expression of CBFα2 (28) and to the low level of CBFα3 induced. In conclusion, the ability of the B cells to express GL α transcripts correlates with the levels of CBFα3 induced by TGF-β1 and the TβRE binding activities.

To assess the function of CBF binding to the GL α promoter, we examined the effects of mutations in the TβRE on reporter gene activity. The GL α promoter −130/+14 or −132/+14 segments containing the wild-type or mutated TβRE (m1 and m2) were cloned upstream of the luciferase reporter gene in the promoterless pXP2 plasmid and were transiently transfected into I.29μ B cells. Consistent with previous data (10), TGF-β1 treatment increased luciferase activity of the plasmid driven by the wild-type GL α promoter by 3.5-fold (Fig. 6). This is in accord with the finding that TGF-β1 increases transcription of endogenous GL α RNA in I.29μ cells by 5-fold in a nuclear run-on assay (8) and also correlates with the 4-fold induction of the TβRE binding activity by TGF-β1 in I.29μ cells (Fig. 1).

The two mutations of the TβRE have different effects on promoter activity. The m1 mutation eliminates TGF-β1 induction and has almost no effect on basal activity. The m2 mutation, which eliminates CBF binding to the TβRE, only reduces TGF-β1 induction by about 30%, but also reduces basal promoter activity by about 50% (Fig. 6). Both mutations reduce the transcriptional activity of the promoter in the presence of TGF-β1.

These transfection data indicate that CBF contributes to both basal activity and to TGF-β1 induction of the GL α promoter. Failure of the m2-TβRE to eliminate TGF-β1 induction of the promoter may be due to the presence of additional CBF sites in the GL α promoter, as demonstrated below. In addition, elimination of TGF-β1 induction of the promoter by the m1 mutation suggests that a protein(s) binding the nucleotides mutated in the m1-TβRE is required for TGF-β1 induction of the promoter. Consistent with this hypothesis are the EMSA data, which indicate that sequence elements mutated in the m1-TβRE are required for optimal binding of the TGF-β1-inducible complex to the TβRE (Fig. 2). Together the data suggest that CBF interacts with another protein(s) that binds the sites defined by the m1 mutation to cause TGF-β1 induction of the GL α promoter.

To directly examine the ability of CBF proteins to activate transcription from the GL α promoter, we cotransfected expression plasmids for CBF proteins and the reporter plasmid driven by the GL α promoter −130/+14 segment into A20.3 cells. We chose A20.3, since these cells express only low levels of CBFα and TβRE binding activity, which are not further induced by TGF-β1 (Fig. 5). Although only CBFα3 is induced in I.29μ cells, we tested the effects of overexpression of all three α subunits on the promoter activity, as it has been shown that all α subunits can trans-activate the TCR β or murine leukemia virus enhancers in vivo (16, 19, 24, 25, 26, 53).

Overexpression of CBFα proteins in A20.3 cells increased the GL α promoter activity by 3- to 7-fold in the absence of TGF-β1 (Fig. 7 A). The response to CBF was increased to 5- to 12-fold in the presence of TGF-β1. The increased response to CBF in the presence of TGF-β1 could be due to post-translational modification of the overexpressed CBFα proteins and/or to interaction of CBF with another TGF-β1-inducible protein(s) in A20.3 cells.

Although CBFβ cannot bind DNA, it heterodimerizes with CBFα proteins and increases their affinity for DNA (16, 17, 18, 25). To determine whether overexpression of CBFβ together with CBFα protein activates the promoter better than CBFα alone, an expression plasmid containing mouse CBFβ cDNA alone or together with a plasmid containing CBFα2 or CBFα3 cDNA was cotransfected into A20.3 cells. Overexpression of CBFβ alone or together with CBFα2 or CBFα3 resulted in luciferase activity marginally greater than that of the pcDNA I vector, CBFα2, or CBFα3 alone, respectively (data not shown). These results indicate that CBFβ is not limiting, consistent with a previous report (15).

The results presented above demonstrate that all CBFα subunits activate the GL α promoter. To determine whether the CBF proteins are acting on the TβRE of the GL α promoter, we tested the effects of TβRE mutations m1 and m2 on activation of the GL α promoter by overexpressed CBF. Since only CBFα3 is induced by TGF-β1 in I.29 μ cells, we only present the effect of overexpression of CBFα3. Overexpression of CBFα2 produced similar results (data not shown).

Cotransfection of an expression plasmid for CBFα3 in A20.3 cells increased activity of the GL α promoter by 5.2-fold, and this activity was reduced about 35% by the m2 mutation in the TβRE (Fig. 7,B), similar to the effect of this mutation on induction by TGF-β1 in I.29μ cells (Fig. 6). The response to CBF was reduced, but again was not eliminated, by the m1 mutation in the TβRE (Fig. 7,B), although this mutation eliminated the response to TGF-β1 (Fig. 6). This indicates that the protein(s) binding to the nucleotides mutated in m1 cannot interact with CBF except in cells treated with TGF-β1. Finally, a DNA segment (−98/+14) that lacked the entire TβRE and adjacent ATF/CRE site (10) had a reduced basal activity but responded well to the overexpression of CBFα3 (only a 20% reduction with respect to the fold induction). These data suggest that other sequence elements in the α promoter must be involved in its activation by CBF.

Failure of the m2 mutation to eliminate TGF-β1 induction of the GL α promoter in I.29μ cells or the response to CBFα3 in A20.3 cells may result from additional CBF binding sites in the GL α promoter. Visual inspection of the sequence within the −130/+14 region revealed an incomplete CBF binding site CCACA at −85/−81 (Fig. 8). We examined the binding activity and TGF-β1 inducibility of complexes formed with a probe (α −95/−78) that contains this element. As shown in Fig. 8,A, two complexes bound this element, one of which was inducible by TGF-β1 (indicated by an arrow), although the binding activity of this probe was much less than that of the TβRE. Competition analyses indicated that the TGF-β1-inducible complex could be competed by a CBF binding site and to a lesser extent by the TβRE and the unlabeled probe itself (Fig. 8,B). Similarly, complexes formed with the TβRE could be inhibited by DNA segments including the −85/−81 CCACA element, but were poorly inhibited by one that does not (−71/+14; Fig. 8 C). Thus, there appears to be another CBF binding site in the GL α promoter in addition to the two sites within the TβRE. This may explain why the m2-TβRE mutation cannot completely eliminate the CBF-response in A20.3 cells.

TGF-β1 is a pleiotropic cytokine, with suppressive effects on growth and both inhibitory and stimulatory effects on the differentiation of a variety of cell types, including B cells. Several DNA sequence elements responsive to TGF-β1 have been identified in a number of genes. These elements include several known transcription factor binding sites, e.g., AP-1, NF-1, USF, and Sp1 sites (54, 55, 56, 57, 58, 59, 60).

In this report we identify two CBF binding sites within the TβRE of the promoter for GL α RNA. We demonstrate that CBF binds to the TβRE and that CBFα3 (AML2) is induced by TGF-β1 in splenic B cells and in two B cell lines, I.29μ and CH12.LX. All three CBFα subunits can activate the GL α promoter, and mutation of the CBF binding sites in the TβRE has similar effects on activation of the promoter by CBFα2 or CBFα3. Although the mechanism of transcriptional activation by CBF is unknown, it has recently been demonstrated that CBFα2 (AML-1B) colocalizes with active RNA polymerase II at the nuclear matrix (61).

Despite their similar trans-activation abilities, the transcriptional targets of the three CBFα proteins in vivo differ due to their variable expression among different tissues and cell lines. Not surprisingly, CBFα genes are regulated by different factors. CBFα1 is induced in fibroblasts by treatment with bone morphogenic protein-7, a member of the TGF-β superfamily. Subsequent to induction of CBFα1, the fibroblasts undergo osteoblast differentiation (29). The trans-activating form of CBFα2 (AML1B) is up-regulated in the embryonic carcinoma cell line F9 cells by retinoic acid (62) and is also increased following denervation of rat skeletal muscle (63). In the present study we demonstrate for the first time that a CBF protein is induced by TGF-β1.

In addition to being induced by TGF-β1, we found that the constitutive levels of CBFα3 appear to correlate with the propensity of the B cells to switch to IgA. Abundant CBFα3 is present in the two B cell lines capable of expressing endogenous GL α transcripts and switching to IgA. In contrast, CBFα3 cannot be induced by TGF-β1 treatment of the sIgG2a⁺ A20.3 B cell line, which also cannot be induced to express endogenous GLα transcripts. The lack of induction of CBFα3 is not due to an inability of A20.3 cells to respond to TGF-β1, as the activity of the GL α promoter is increased in cells treated by TGF-β1 (Fig. 7 A) (10).

It is possible that the constitutive expression of CBFα3 may be due to the autocrine TGF-β1 that is produced by I.29μ cells. It is unknown whether CH12.LX cells secrete TGF-β1. Although CBFα3 is not detectable in splenic B cells in the absence of TGF-β1, it is inducible by treatment with LPS and TGF-β1. Splenic B cells have the ability to switch to all Ig classes depending on the cytokine and activation signal received. B cells are believed to become committed to switch to IgA in Peyer’s patches, where intestinal Ags are presented to B and T cells. Furthermore, GL α transcripts are induced in Peyer’s patches during the initiation of an immune response (64). It is possible that CBFα3 may be induced in B cells in Peyer’s patches in the presence of activation signals and that this contributes to initiation of transcription of GL α RNA.

Regulation of a promoter requires binding of multiple factors. Previous studies of the human GL α1 and mouse GL α promoters indicate that the TβRE, ATF, and Ets sites are all important for induction of these promoters by TGF-β1 (10, 12) (M.-J. Shi, unpublished observations). In this study we provide data indicating that CBF along with another unidentified transcription factor bind to the TβRE and that binding of both factors is required for TGF-β1 induction. Since the m1 mutation eliminates the response of the GL α promoter to TGF-β1, but not that to overexpressed CBF, it is possible that the putative protein(s) binding the sites mutated in m1 is activated by TGF-β1.

CBFα has been shown to interact with Ets proteins and to synergistically activate promoters binding both proteins (21, 53, 65, 66). Consistent with this, we found that simultaneous mutation of the CBF sites in the TβRE and of the Ets site in the GL α promoter greatly reduces the activity of the promoter (90%) and the response to overexpressed CBF (50%). Mutation of either the CBF or Ets site alone has much less of an effect on the response to CBF (data not shown). These results suggest that Ets and CBF may cooperate to activate the GL α promoter.

Members of a recently identified family of proteins, the Smads, serve both as signal transducers and transcriptional activators for signaling by the TGF-β superfamily (reviewed in Refs. 67 and 68). Upon TGF-β1 binding to its receptors, Smad2 and Smad3 are rapidly phosphorylated and form a complex with Smad4 (DPC4). This complex is translocated into the nucleus and binds DNA. Although natural binding sites in promoters that respond to Smads are not well characterized, optimal binding sites for Smad3 and Smad4 were recently identified (69). They closely resemble the repeated sequence elements mutated in the m1 mutation. It will be interesting to investigate the relationship between activation of Smads and induction of CBFα3 proteins by TGF-β1 and to determine whether Smad proteins are involved in the regulation of GL α transcription in B cells.

We are especially grateful to Dr. J. Stein (University of Massachusetts Medical School, Worcester, MA) for pointing out the similarity between the CBF binding site and the TβRE. We also thank Drs. J. Stein and C. Banerjee for their generosity with Abs and for helpful suggestions. We thank Dr. S. Hiebert (National Cancer Institute, Frederick, MD) for CBFα1, -2, and -3 and CBFβ Abs and CBFα expression plasmids. We thank Dr. N. Speck (Dartmouth Medical School, Hanover, NH) for helpful suggestions and for CBFα2 and CBFβ expression plasmids, recombinant CBFα2 (Runt domain), and CBFβ Ab. We are also grateful to our colleagues Drs. C. Schrader and F. He and Mr. C. H. Shen for their helpful comments on the manuscript.