The B cell lymphoma-6 (BCL-6) transcriptional repressor protein is an important regulator of B cell differentiation and is strongly implicated in the development of B cell lymphoma. Expression of the Blimp-1 transcription factor, which is critical for promoting B cell differentiation into plasma cells, is repressed by BCL-6. We have investigated the mechanism for how BCL-6 represses Blimp-1 transcription, and have found that BCL-6 regulates the Blimp-1 promoter through a novel mechanism involving AP-1 elements. Specifically, BCL-6 is a potent repressor of transcriptional activity mediated by AP-1 factors. We found that the zinc-finger region of BCL-6 interacts with c-Jun, JunB, and JunD proteins but does not bind c-Fos or Fra-2 proteins. An estrogen receptor ligand binding domain fusion with the BCL-6 zinc finger domain can act as a estrogen-inducible dominant negative protein and increase AP-1 activity in BCL-6+ cells but not in BCL-6− cells, indicating that endogenous BCL-6 represses AP-1 activity. Additionally, we have confirmed a specific interaction between c-Jun and the zinc finger domain of BCL-6 in vivo using a mammalian two-hybrid assay. Repression of AP-1 function by BCL-6 may be a key mechanism for how BCL-6 regulates gene expression to control inflammation, lymphocyte differentiation, and lymphomagenesis.
The differentiation of naive B cells into Ab-secreting cells is a central component of the Ag-specific immune response. Recent insights into this process have been obtained by analyzing the regulation of gene expression by specific transcription factors during B cell differentiation. Two important transcription factors that are regulated during B cell differentiation and control different aspects of B cell differentiation are B cell lymphoma-6 (BCL-6)3 and Blimp-1 (1, 2). BCL-6 was initially identified as a gene that is a frequent target of chromosomal translocations in diffuse large cell lymphoma (DLCL), a common subtype of non-Hodgkin’s B cell lymphoma (reviewed in Refs. 3 and 4). BCL-6 encodes a transcriptional repressor protein containing an N-terminal pox virus and zinc finger (ZF) domain (POZ) and six ZF. BCL-6 is expressed at the highest levels in B cells in the germinal center reaction and is down-regulated upon B cell differentiation into plasma cells. Mice deficient in BCL-6 develop normal primary Ab responses but fail to develop germinal centers, indicating that BCL-6 is not required for plasma cell differentiation but is essential for the transition of B cells into the germinal center stage (5, 6, 7). DLCL B cells that express BCL-6 also have a germinal center B cell phenotype (8), and alterations in BCL-6 expression in DLCL may prevent down-regulation of BCL-6 and thus prevent terminal B cell differentiation into a plasma cell. One explanation for the inhibitory effect of BCL-6 on B cell differentiation comes from recent studies that indicate that BCL-6 can repress the transcription of the Blimp-1 gene (9, 10). Blimp-1 is a transcriptional repressor protein induced during differentiation into plasma cells, and is not expressed in BCL-6 positive germinal center B cells (1, 2). Forced expression of Blimp-1 can induce the terminal differentiation of B cells into plasma cells (10, 11). Blimp-1 is capable of repressing c-myc transcription, and thus Blimp-1 may be critical for the nonproliferative state of normal plasma cells (12). Accordingly, Blimp-1 is a putative tumor suppressor gene and the chromosomal region encoding Blimp-1 is frequently deleted in lymphoma (13). Therefore, a crucial role for BCL-6 may be to repress the expression of Blimp-1 during B cell differentiation. This repression of Blimp-1 expression by BCL-6 may be exploited particularly by tumor cells in DLCL, however it remains to be determined whether Blimp-1 is the major target for the B cell transformation function or whether other BCL-6 target genes play critical roles.
Many questions regarding the basic mechanisms for how BCL-6 represses target genes are currently unanswered. For instance, no BCL-6 target promoter has been analyzed for how BCL-6 repression affects other transcription factors that positively regulate promoter activity and how these effects integrate together. Although BCL-6 binds to corepressor proteins that recruit histone deacetylases (14, 15, 16), the precise role of these accessory proteins for repressing natural target promoters has not been determined. Additionally, a major question that remains unanswered is how BCL-6 represses target genes that do not have clear BCL-6 DNA binding sites in their promoters. Thus, BCL-6 may regulate expression of some target genes by either acting through other transcription factors or by binding to distal regulatory elements away from the promoter. Finally, an important aspect of understanding how BCL-6 regulates B cell differentiation is determining the mechanism for how BCL-6 regulates Blimp-1 transcription. In this study, we have analyzed the molecular mechanism by which BCL-6 represses the Blimp-1 promoter. In the course of this analysis, we uncovered a novel regulatory pathway by which BCL-6 can inhibit AP-1 transcriptional activity. We found that the ZF region of BCL-6 can specifically bind to Jun proteins (c-Jun, JunB, and JunD) that make up AP-1, and that the isolated ZF domain of BCL-6 can modulate AP-1 transcription. Increased AP-1 activity in the absence of repression by BCL-6 can explain the lack of germinal center B cells in BCL-6-deficient mice, as well as other immune abnormalities in these mice. Finally, repression of Blimp-1 expression by a BCL-6-AP-1-dependent mechanism may represent a key control point for both B cell differentiation and B cell malignant transformation.
Materials and Methods
Cell lines and reagents
Raji B lymphoma and ARH77 myeloma cells were maintained in RPMI 1640 medium supplemented with 10% FCS plus antibiotics. NIH3T3 cells were maintained in DMEM supplemented with 10% FCS plus antibiotics. The Raji-ZnFERD cell line was prepared as previously described (9) and maintained in phenol red-free RPMI 1640 medium plus 10% charcoal-treated FCS and antibiotics. PMA, ionomycin, and β-estradiol were obtained from Sigma-Aldrich (St. Louis, MO).
Reporter and expression constructs
The VXY-ZnFERD retroviral construct (9) was obtained from Dr. A. Shaffer (National Cancer Institute, National Institutes of Health, Bethesda, MD). A 2000-bp region of the human Blimp-1 promoter was isolated from human (Raji cell) genomic DNA by PCR using 5′-ATTTTAGGGTTGCCTGGCTAAGCC-3′ as the forward and 5′-GTCACGGCAGCACTTTGTCTGTGT-3′ as the reverse primers. The PCR product obtained was ligated into the pcR2.1 TOPO cloning vector (Invitrogen, Carlsbad, CA) and then subcloned into pGL3-basic (Promega, Madison, WI) using SacI and XhoI sites. The promoter construct was sequenced to verify that the proper sequence was amplified. The mutant AP-1 site promoter, Mut.Blimp-1pGL3 was constructed by replacing the two AP-1 sites at −1813 and −1647 with BamHI and MluI restriction enzyme sites respectively. The −1813 AP-1 site was mutated by amplification of the promoter with the 5′-ATTTTAGGGTTGCCTGGCTAAGCC-3′ as forward primer and 5′-GGATCCACCATGATAGAGTAAGGGAATAGG-3′ as the reverse primer. The −1647 AP-1 site was mutated by amplification of the −1813 AP-1 mutated promoter with 5′-AAGGACGCGTGCATCAGTTAATAAACCCTCTTGA-3′ as the forward and 5′-GTCACGGCAGCACTTTGTCTGTGT-3′ as the reverse primers. The mutant promoter was cloned into pGL3-basic via KpnI and XhoI restriction sites. The 3× AP-1 reporter contains three canonical AP-1 binding sites (TGACTCA) upstream of a minimal promoter fragment containing a TATA box in the luciferase reporter plasmid pGL3-basic. The Rous sarcoma virus (RSV) promoter reporter was constructed by isolating the RSV promoter from pRc/RSV (Invitrogen) via BglII and HindIII sites and cloned into pGL3-basic by the same sites. CXN-BCL-6 was constructed by inserting the human BCL-6 cDNA from pBS-BCL-6 full length (FL) (17) via SalI ends into the XhoI site in the pCXN2 expression vector. The pCXN2 (CXN) vector was obtained from Dr. K. Ozato (National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, MD) and consists of the human BCL-6 cDNA driven by a hybrid β-actin-CMV promoter. Constructs for c-Jun, JunB, JunD, c-Fos, Fra-1, and Fra-2 were the kind gift of Dr. M. Sadar (Department of Cancer Endocrinology, British Columbia Cancer Agency, Vancouver, Canada) with the permission of Dr. F. Saatcioglu (Department of Pharmacology, School of Medicine, University of California, San Diego, CA). The CXN-BCL-6 ZF deletion was made by digesting pBS-BCL-6FL with AspI, which cleaves a unique site in the human BCL-6 cDNA 45 bp before the start of the ZF domain, and HindIII, which cleaves the pBS vector. The digested plasmid was separated from the smaller deleted fragment, and ligated. The truncated cDNA was then cloned into pCXN2 via SalI and XhoI.
Transfections and luciferase assays
Luciferase assays were performed as previously described (18). Transfections of NIH3T3 cells were performed using Fugene (Roche, Indianapolis, IN) as a transfection reagent similar to that described previously (18). A standard transfection reaction used 2 μg of DNA, with 1 μg of reporter construct, 0.5 μg of RSV expression construct (either RSV alone or RSV-c-Jun plus c-Fos), and 0.5 μg of either CXN alone or CXN-BCL-6. The experiments presented show representative luciferase data from multiple experiments. For a standard assay, two transfection mixes were each tested in duplicate and the average is shown. Raji and ARH77 cells were transfected by electroporation using a Bio-Rad GenePulser set at 280 mV, 960 μF. Cells were suspended in serum-free RPMI 1640 medium at 1.25 × 106 per ml, and placed in a 0.4-cm cuvette with 15 μg of reporter construct. After electroporation, cells were resuspended in RPMI 1640 plus 10% FCS, with or without β-estradiol treatment (5 μM final concentration). Luciferase assays were performed 24 h after electroporation. Raji and ARH77 cells were stimulated with PMA plus ionomycin for 6 h before harvesting for luciferase, except where indicated. PMA was used at 20 ng/ml and ionomycin was used at 0.3 μM.
GST-fusion protein binding assays
GST-fusion protein preparation and GST binding assays were performed as described previously (19). Briefly, equivalent of amounts of beads with either GST protein alone or GST-fusion protein were loaded into tubes as a 1:1 slurry with washing buffer. Glutathione-Sepharose beads were used to bring the bead volume up to 25 μl in all instances. The GST alone and GST-fusion protein beads were combined with 5 μl of either [35S]methionine- or [35S]cysteine-labeled in vitro translation reactions in a final volume of 500 μl interaction buffer (washing buffer + 10% glycerol) on a rotary shaker for 90 min at 4°C. For the nuclear extract pull-down, 500 μg of nuclear protein was used for each binding reaction. Nuclear proteins were prepared as described previously (18). The beads were washed five times with two volumes of washing buffer. The bound proteins were eluted, subjected to SDS-PAGE, and visualized by autoradiography in the case of translated proteins, and by immunoblot in the case of nuclear extract. Immunoblots were probed with the N-3 anti-BCL-6 Ab (Santa Cruz Biotechnology, Santa Cruz, CA). In vitro translated proteins were made with a coupled T3/T7-polymerase transcription and rabbit reticulocyte lysate translation system (Promega). FL and POZ minus (PM) forms of BCL-6 in the pBS plasmid have been described (17). The middle region form of BCL-6 was prepared by digesting pBS-BCL-6PM with SphI, which recognizes a unique site in the human BCL-6 cDNA just upstream of the region coding for the ZF. This linearized plasmid was then used for in vitro translation. The ZF alone region of BCL-6 was generated by PCR using pBS-BCL-6FL as a template with the following oligos: 5′-GAC CTG GGA TCC ACC ATG TGT GAG AAC GGG GCC TTC TTC TGC-3′ and 5′-AGG TTC CTC GAG ACT AGT TCA CTT GTC GTC ATC GTC CTT GTA-3′. The PCR product was cloned into pBS after BamHI and XhoI digestion, and then used for in vitro translation as above. The plasmid encoding GST-c-Jun was a kind gift of Dr. H. Nakshatri (Department of Surgery, Indiana University School of Medicine, Indianapolis, IN). GST-BCL-6-ZF (GST-ZF) was generated by PCR using pBS-BCL-6FL as a template with the following oligos: 5′-AAAAAAGGATCCACCATGGGCCCCACGTTCGCTGAGGAGATGGGA-3′ and 5′-AAAAAAGGATTCTCACTTGTCGTCATCGTC-3′. The PCR product was cloned into pGEX-2tk (Amersham Pharmacia Biotech, Piscataway, NJ) after BamHI and EcoRI digestion, and then used for production of GST-fusion protein. The following pBS constructs were made for in vitro translation of c-Jun, c-Fos, JunB, JunD, and Fra-2. The cDNA for c-Jun was amplified from RSV-c-Jun plasmid by PCR using the primers 5- AAAAAAGGATCCATGACTGCAAGATGGAAAC-3′ and 5′-AAAAAAGAATTCTCAAAATGTTTGCAACTGCT-3′. The PCR product was digested with BamHI and EcoRI and cloned into pBS. The cDNA for c-Fos was amplified from RSV-c-Fos plasmid by PCR using the primers 5′-ATGATCTTCTCGGGCTTCAACGCAGAC-3′ and 5′-TCACAGGGCCAGCAGCGTGGGTGAGCT-3′. The c-Fos PCR product was cloned into the vector pCR2.1TOPO (Invitrogen), re-isolated by digestion with EcoRI, and cloned into pBS. The cDNA for Jun-B was amplified from RSV-JunB by PCR using the primers 5′-AAAAAAGAATTCATGTGCACGAAAATGGAACAGCCT-3′ and 5′-AAAAAAGAATTCTCAGAAGGCGTGTCCCTTGACCCCTAG-3′. The PCR product obtained was digested with EcoRI restriction enzyme and cloned into pBS. RSV-JunD was digested with HindIII and EcoRI to isolate JunD cDNA, which was then cloned into pBS. RSV-Fra2 was digested with HindIII and NotI to isolate the JunD cDNA, which was then cloned into pBS.
Mammalian two-hybrid assay
The Matchmaker Mammalian Two-Hybrid kit was obtained from Clontech Laboratories (Palo Alto, CA), and the plasmids provided were used for further cloning. The pM plasmid was used to produce a c-Jun-Gal4 DNA binding domain fusion protein-encoding plasmid. The pVP16 plasmid was used to produce a BCL-6 ZF domain-VP16 fusion protein-encoding plasmid. PCR products for c-Jun and BCL-6 ZF (BZF) domains were cloned into pM and pVP16 similarly as cloning into pGEX2tk described in the previous section. The Gal4 responsive promoter from pG5CAT was cloned into the pgl3 luciferase reporter plasmid to use luciferase as a read-out for the two-hybrid assay. The Gal4 responsive promoter was first PCR amplified using the primers 5′-AAAAAACTCGAGACCATGATTACGCGCCAAGCTAAT-3′ and 5′-AAAAAAAAGCTTTGATGAATTCGAGCTGGCGCATTA-3′ and after digestion with XhoI and HindIII, cloned into the pgl3 luciferase plasmid via the same sites. Two hybrid assays were performed by transfecting and harvesting NIH3T3 cells as described (18). Typical transfection mixes contained a total of 2 μg DNA with 0.2 μg pgl3-Gal4 plasmid, 0.9 μg pM or pM-c-Jun, and 0.9 μg pVP16 or pVP16-BZF.
The human Blimp-1 promoter is regulated by BCL-6
An initial analysis of the sequence of the human Blimp-1 promoter revealed no clear BCL-6 binding sites. A sequence with similarity to a BCL-6 binding site and a STAT factor binding site was found at −1309 with the sequence TTCTCTGAA (Fig. 1⇓A, STAT-like). Although this −1309 sequence is not an optimal BCL-6 binding site (TTC [C/T] [T/A] [G/A] GAA), we tested this site for binding by BCL-6 by gel shift analysis and detected only extremely weak binding compared with a probe containing an optimal BCL-6 binding site (data not shown). Because BCL-6 might still bind to this −1309 STAT-like site in vivo, we tested BCL-6 binding to this region by chromatin immunoprecipitation. Using chromatin immunoprecipitation, we detected BCL-6 binding to a region from the CD69 promoter that contains an optimal BCL-6 binding site, but we did not detect BCL-6 binding to the −1309 sequence (data not shown).
We decided first to test whether BCL-6 could specifically repress the Blimp-1 promoter, because BCL-6 might regulate Blimp-1 transcription through elements outside of the promoter. Therefore, we reconstructed the Raji-ZnFERD cell line (9), which expresses the BCL-6 ZF (ZnF) domain as a fusion protein with the ligand-binding domain of the estrogen receptor (ERD). In this cell line, β-estradiol stimulation causes translocation of the ZnFERD protein into the nucleus, where the ZF fusion protein can compete for binding with the ZF region of endogenous BCL-6 expressed in these cells. Previous experiments showed that β-estradiol induction of the ZnFERD protein can lead to an increase in endogenous Blimp-1 expression, as well as multiple other changes in gene expression consistent with derepression of putative BCL-6 target genes (9). We therefore tested the ability of the ZnFERD protein to activate the Blimp-1 promoter (Fig. 1⇑B). Raji-ZnFERD cells transfected with the Blimp-1 promoter reporter and then stimulated with PMA plus ionomycin showed a significant increase in luciferase activity after β-estradiol treatment. This data indicated that BCL-6 regulates Blimp-1 promoter itself, and does not require distal regulatory elements for its action. However β-estradiol only increased Blimp-1 promoter activity when Raji-ZnFERD cells were treated with PMA plus ionomycin (Fig. 1⇑B). This indicates that the Blimp-1 promoter is responsive to PMA plus ionomycin stimulation. This also suggests that BCL-6 might require transcription factors regulated by PMA plus ionomycin to repress the Blimp-1 promoter. Therefore, we investigated the precise elements in the Blimp-1 promoter that are regulated by BCL-6. At the distal end of the 2-kb region of the Blimp-1 promoter are two canonical AP-1 binding sites (Fig. 1⇑A). Because Blimp-1 activity was only affected by the BCL-6 ZF after PMA plus ionomycin stimulation of Raji-ZnFERD cells, and it is known that PMA plus ionomycin strongly activate AP-1 proteins in lymphoid cells, we decided to test whether BCL-6 might regulate transcription through AP-1 proteins.
Repression of AP-1 activity by BCL-6
We tested the hypothesis that BCL-6 might regulate the Blimp-1 promoter through AP-1 elements by mutating both the consensus AP-1 sites in the Blimp-1 promoter. We then assayed the ability of BCL-6 to repress either the wild-type Blimp-1 promoter or the double AP-1 sites mutated (Mut.Blimp-1) promoter. We used a transient transfection assay with both NIH3T3 mouse fibroblasts and PMA plus ionomycin stimulated ARH77 human myeloma cells (Fig. 2⇓A). NIH3T3 cells grown in 10% FCS express c-Jun, and have significant levels of AP-1 activity in the absence of added stimulus (20). ARH77 cells are BCL-6 negative and constitutively express Blimp-1; PMA plus ionomycin stimulation strongly up-regulates c-Fos and c-Jun in these cells and weakly up-regulates Blimp-1 (data not shown). The wild type and mutated forms of the Blimp-1 promoter produced strong luciferase activity in both ARH77 cells and NIH3T3 cells, although in NIH3T3 cells the mutated promoter had 4-fold lower activity compared with the wild type promoter. These results indicate that the Blimp-1 AP-1 sites are functional but are not essential for overall activity of the promoter.
We next determined that the wild-type Blimp-1 promoter was repressed from 2.4- to 4.8-fold by BCL-6 in ARH77 and NIH3T3 cells, respectively (Fig. 2⇑A). In contrast, the mutated Blimp-1 promoter was not significantly repressed by BCL-6 in either cell type (Fig. 2⇑A). Thus, mutation of AP-1 binding sites essentially removed the ability of BCL-6 to repress the Blimp-1 promoter by BCL-6 in NIH3T3 cells. These experiments were also repeated in the human 293 cell line with similar results, showing that this result is not restricted to selected cell lines (data not shown). These results indicated that BCL-6 acts through AP-1 elements to regulate the Blimp-1 promoter, and these results are consistent with the data in Fig. 1⇑B showing that the ZF of BCL-6 could activate the Blimp-1 promoter in the absence of a clear BCL-6 binding site.
To further test the hypothesis that BCL-6 regulates AP-1 activity, we constructed a luciferase construct with a minimal promoter driven by three multimerized consensus AP-1 sites (3× AP-1). As a control for repression specificity, we tested a luciferase construct driven by the RSV promoter. We again assayed for the ability of BCL-6 to repress the two promoters using a transient transfection assay in NIH3T3 cells (Fig. 2⇑B). Whereas the RSV promoter was resistant to repression by BCL-6, there was a 3.9-fold repression of the 3× AP-1 promoter by BCL-6. The 3× AP-1 reporter was also repressed BCL-6 in ARH77 cells (Fig. 2⇑B), indicating that BCL-6 could repress AP-1 activity in a general manner apart from the Blimp-1 promoter. We used gel shift assays to rule out the possibility that BCL-6 could bind directly to a consensus AP-1 site (data not shown). Furthermore, chromatin immunoprecipitation experiments showed that BCL-6 did not bind to the AP-1 sites of the endogenous Blimp-1 promoter (data not shown), indicating that in vivo BCL-6 does not bind to AP-1 binding sites. Thus, BCL-6 appears to repress AP-1 transcriptional activity by a mechanism apart from binding to AP-1 DNA binding sites.
Physical interaction between BCL-6 and Jun proteins
To determine how BCL-6 specifically repressed AP-1 activity, we next investigated whether BCL-6 and AP-1 proteins interact by using a GST pull-down assay. We started by testing whether GST-c-Jun could specifically interact with BCL-6 (Fig. 3⇓, lanes 1–15). GST-c-Jun is very effective at binding to BCL-6 present in Raji nuclear extracts (Fig. 3⇓, lanes 1–3), indicating BCL-6 can effectively compete for c-Jun binding in the presence of other nuclear proteins. Using in vitro translated proteins, we found that GST-c-Jun could reproducibly interact with both FL BCL-6 protein, and a mutant BCL-6 PM protein missing the N-terminal POZ domain, whereas these forms of BCL-6 had marginal affinity for GST protein alone (Fig. 3⇓, lanes 4–9). The interaction of BCL-6 with c-Jun is unusual therefore, because it is independent of the POZ domain, and other proteins that have been reported to interact with BCL-6 require the BCL-6 POZ domain. We next determined that while c-Jun does bind to the middle region of the BCL-6 protein, the BCL-6 ZF region shows a high affinity for c-Jun (Fig. 3⇓, lanes 10–15).
We next tested the ability of the ZF region of BCL-6 expressed as a GST-fusion protein to bind c-Jun (Fig. 3⇑, lane 17). When we analyzed the specificity of AP-1 protein binding to GST-ZF, we found that all three Jun proteins showed a high degree of interaction with BCL-6, whereas c-Fos and Fra-2 did not interact with the GST-ZF (Fig. 3⇑, lanes 18–30). Although BCL-6 does not bind directly to c-Fos, BCL-6 has the potential to inhibit c-Fos-driven AP-1 activation because c-Fos only binds to DNA as a heterodimer with c-Jun, JunB, or JunD (21). Thus BCL-6 protein binding to Jun proteins can provide a mechanism for the inhibition of AP-1 transcriptional activity by BCL-6.
Activation of AP-1 transcription by the BCL-6 ZF
We next wondered whether the ZF domain of BCL-6 could act as a dominant negative protein to block transcriptional repression by full-length BCL-6. We therefore tested whether Raji-ZnFERD cells had alterations in AP-1 activity after β-estradiol induction (Fig. 4⇓). Although untransfected control Raji cells showed a minor activation of AP-1 transcription after β-estradiol treatment, Raji-ZnFERD cells treated with β-estradiol showed an almost four-fold increase in AP-1 activity compared with untreated cells. Thus, nuclear induction of the ZnFERD protein results in AP-1 activation in BCL-6 positive Raji cells. In contrast, BCL-6-negative ARH77 cells transfected with the ZnFERD construct actually showed a decrease in AP-1 activity after β-estradiol treatment (Fig. 4⇓B), indicating that the ZnFERD protein is not a general activator of AP-1 transcription, but actually is an AP-1 inhibitor by itself. The repression seen with the ZnFERD protein in ARH77 cells may be explained by recent results from Lemercier et al. (22) in which the ZF domain of BCL-6 was shown to contain a novel transcriptional repression domain capable of binding histone deacteylases. Although the ZF domain of BCL-6 can repress transcription autonomously, full-length BCL-6 is a more potent repressor of transcription (22) (L. M. Toney and A. L. Dent, data not shown). The results in Fig. 4⇓ indicate that the endogenous full-length BCL-6 in Raji cells is a strong transcriptional repressor of AP-1 activity, and that the ZnFERD protein can inhibit a significant amount of this repression. Thus, although the ZnFERD protein may repress transcription on its own, in competition with full-length BCL-6, expression of the ZnFERD protein results in relative transcriptional activation.
To further test the importance of the BCL-6 ZF domain in repressing AP-1 activity, we constructed a BCL-6 protein that has a normal N-terminal 500 amino acids, but is missing the entire C-terminal ZF domain. This protein was expressed in NIH3T3 cells at levels comparable to full-length BCL-6 (data not shown). However, when tested by transient transfection assay in NIH3T3 cells, the ZF deleted BCL-6 protein had a dramatically decreased ability to repress c-Jun plus c-Fos activated AP-1 activity (Fig. 4⇑C). These data therefore indicate that the ZF domain of BCL-6 is critical for the repression of AP-1 transcription, and further these data support the idea that the BCL-6 ZF domain can bind to c-Jun in cells and repress AP-1 activity.
Finally, we wanted to test further the interaction between the BCL-6 ZF domain and c-Jun by using a mammalian two-hybrid system. In this assay, we constructed two fusion proteins, c-Jun fused to the Gal4 DNA binding domain (pM-c-Jun) and the BCL-6 ZF domain fused to the VP16 transcriptional activator (pVP16-BZF). When tested in NIH3T3 cells with a Gal4 responsive reporter plasmid, as expected, pM-c-Jun activated the reporter compared with pM alone (Fig. 5⇓). However, the transcriptional activity seen with pM-c-Jun plus pVP16-BZF was markedly higher (4.5-fold over pM-c-Jun alone) by pVP16-BZF (Fig. 5⇓). These data strongly support the idea that the ZF region of BCL-6 can bind to c-Jun in cells and modulate transcriptional activity. The strength of interaction between the BCL-6 ZF domain and c-Jun is difficult to measure in the two-hybrid assay, because the assay relies on transcriptional activation, and the ZF region has some degree transcriptional repression function. The fact that we were able to measure some degree of interaction in this assay indicates that the VP16 activation domain can overcome the repression mediated by the ZF domain. An important point is that this mammalian two-hybrid assay probably under-represents the degree of interaction between c-Jun and the BCL-6 ZF domain in vivo, because the transcriptional activation mediated by VP16 is partially offset by transcriptional repression mediated by the BCL-6 ZF domain.
In this study we have characterized transcriptional repression of the Blimp-1 promoter by the BCL-6 protooncogene. Unexpectedly, we have found that BCL-6 represses transcription of the Blimp-1 promoter almost entirely by inhibiting the transcriptional activity of AP-1 proteins. Our data also show a general role for BCL-6 in modulating AP-1 responses, and this information provides a mechanism for the regulation of BCL-6 target genes that do not contain clear BCL-6 DNA binding sites. In fact, many currently identified “BCL-6 target genes” may actually be AP-1 target genes that are indirectly repressed by BCL-6. Independent of how BCL-6 target genes are defined, our results have important implications for the normal biological function of BCL-6. In particular, our data have significance for the regulation of the immune system by BCL-6, as well as the role of BCL-6 in B cell lymphoma.
BCL-6 deficient mice have several striking immunological phenotypes including lack of germinal centers (5, 6, 7). Our findings on inhibition of AP-1 function by BCL-6 relate to the regulation the germinal center reaction by BCL-6. Normal germinal center B cells express very high levels of BCL-6 protein (23, 24). Thus, BCL-6 should repress AP-1 activity very strongly in germinal center B cells, and BCL-6 may play a critical role in suppressing AP-1 activity to maintain the germinal center reaction. Although there is little data on the role of Jun proteins in germinal center B cells, studies of c-Fos transgenic mice have revealed that over-expression of c-Fos inhibits the germinal center reaction (25). Our data predicts that over-expression of c-Fos in mice mimics BCL-6 deficiency, because both types of mice should have increased AP-1 activity. Increased AP-1 activity will lead to Blimp-1 expression at an earlier stage than normal, leading to premature generation of plasma cells and inhibiting germinal center B cell development. B cells from c-Fos transgenic mice also accumulate the cell cycle inhibitor p27kip1 (26), and p27kip1 has been reported to be a target of repression for BCL-6 (9). Increased expression of the cell cycle inhibitor p27kip1 (in c-Fos transgenic mice or BCL-6 deficient mice) may be another factor affecting germinal center formation in these mice. Like Blimp-1, p27kip1 does not have a BCL-6 binding site in its promoter (A. L. Dent, unpublished observations), suggesting that BCL-6 regulates p27kip1 through repression of AP-1 transcriptional activation. Because DLCL is similar in phenotype to germinal center B cells (8), our current insights into the mechanism for how BCL-6 regulates B cell differentiation may be key to developing therapies that can specifically target DLCL.
BCL-6 deficient mice also have a high frequency of Th2-type inflammatory disease of the heart and lung (5, 7). Thus, BCL-6 is essential for normal T cell differentiation as well as inhibiting inflammatory disease. Previous studies have implicated AP-1 proteins in IL-4 transcription and subsequent Th2 cell commitment and function (27, 28, 29, 30). Our data suggests that BCL-6 deficient T cells may have amplified AP-1 activity that can lead to increased IL-4 expression and Th2 differentiation. We have recently reported that BCL-6 plays an important role in macrophage chemokine expression (18), and increased AP-1 activity in BCL-6 deficient macrophages may explain the pro-inflammatory activity of these cells. Moreover, we have found that BCL-6 deficient macrophages over-express several pro-inflammatory genes, and many of these genes do not contain BCL-6 binding sites in their promoters (A. L. Dent and L. M. Toney, unpublished data). Perturbations in AP-1 activity in myocardial cells in the absence of BCL-6 may also provide a mechanism for myocardial cell damage in BCL-6 deficient mice (31). Such cell damage may be an important initiating factor in the severe heart inflammation that strikes the majority of BCL-6 deficient mice. BCL-6 deficiency in regulatory cells such as macrophages and T cells could then amplify the inflammatory response. The Th2 nature of the inflammatory disease may be at least partially explained by increased AP-1 activity in the absence of BCL-6.
We have determined that BCL-6 is capable of binding the AP-1 proteins c-Jun, JunB, and JunD, providing a mechanism of direct binding for the repression of AP-1 activity by BCL-6. Although we have obtained clear data for a BCL-6/c-Jun interaction with GST-pull-down assays and the mammalian two-hybrid assay, we have not been successful in showing binding of BCL-6 to c-Jun by coimmunoprecipitation experiments with either transfected or endogenous proteins (data not shown). Thus, the binding of BCL-6 to Jun proteins may be a relatively low affinity interaction, or a transient interaction, or a complex interaction that requires other proteins to stabilize the interaction in vivo. Regardless of the nature of the interaction, the striking effects of BCL-6 on c-Jun/AP-1 transcriptional function indicate that the physical interaction of these BCL-6 and Jun proteins is biologically relevant. The biological significance of the BCL-6-c-Jun interaction is further strengthened by the fact that Raji-ZnFERD cells can activate AP-1 transcription when treated with β-estradiol. The interpretation of this experiment is that the ZnFERD protein can bind to Jun proteins and disrupt binding of endogenous BCL-6, resulting in AP-1 transcriptional activation. In the absence of endogenous BCL-6, ZnFERD proteins can inhibit AP-1 transcription (Fig. 4⇑B). We have also seen relatively low levels of AP-1 repression with transfecting just the BCL-6 ZF domain (data not shown), suggesting that the BCL-6 ZF domain alone can bind to Jun proteins and inhibit AP-1 activity, possibly by recruiting histone deacetylases (22). Nonetheless, the fact that β-estradiol-induction of the ZnFERD protein in Raji cells results in relative AP-1 activation (Fig. 4⇑A), most likely indicates that endogenous BCL-6 can repress AP-1 activity more effectively than the ZF domain alone.
One possibility for the specific mechanism of repression by BCL-6 is that the ZF region of BCL-6 is able to bind AP-1/Jun proteins and thus bring the potent transcriptional repression activity of BCL-6 to AP-1 complexes bound to DNA. An alternative possibility is that the ZF region of BCL-6 binds to c-Jun and disrupts dimerization of c-Jun with itself or other AP-1 proteins. An inhibition of dimerization of c-Jun by BCL-6 would lead to repression of AP-1 responses because c-Jun complexes would not be able to bind DNA. BCL-6 may also inhibit AP-1 activity through a combination of both mechanisms. Although we think our data supports a mechanism where BCL-6 binds to c-Jun complexes bound to DNA and recruits in corepressor factors, we cannot completely rule out that BCL-6 also inhibits c-Jun dimerization. We have found that while the ZF region of BCL-6 alone can repress AP-1 activity, full-length BCL-6 can repress AP-1 more strongly than the ZF region alone (Fig. 4⇑ and data not shown). Full-length BCL-6 does not seem to bind c-Jun any more effectively than the ZF region of c-Jun (Fig. 3⇑). Thus, full-length BCL-6 may be able to repress AP-1 activity more effectively than the ZF domain due to the greater ability of the POZ domain to recruit corepressor factors (14, 15, 16). Nonetheless, we cannot currently rule out the possibility that in vivo full-length BCL-6 is more effective than the ZF domain alone at disrupting c-Jun dimerization and DNA binding. Investigation of the precise molecular mechanism for repression of AP-1 activity by BCL-6 will be the subject of future research.
Taken together, our results indicate a novel and important role for the BCL-6 protooncogene in repressing AP-1 transcription. This new function for BCL-6 should provide new insights into B cell differentiation and B cell malignant transformation.
We thank Dr. Cheong-Hee Chang for critically reviewing the manuscript and Dr. Marianne Sadar for generously supplying AP-1 plasmids.
↵1 This work was partially supported by V Foundation Scholar and Showalter Trust Awards (to A.L.D.).
↵2 Address correspondence and reprint requests to Dr. Alexander L. Dent, Department of Microbiology and Immunology and The Walther Oncology Center, 1044 West Walnut Street R4 302, Indiana University School of Medicine, Indianapolis, IN 46202. E-mail address:
↵3 Abbreviations used in this paper: BCL-6, B cell lymphoma-6; DLCL, diffuse large cell lymphoma; RSV, Rous sarcoma virus; FL, full length; ZF, zinc finger; ERD, ligand-binding domain of the estrogen receptor; POZ, pox virus and ZF domain; PM, POZ minus.
- Received March 28, 2002.
- Accepted June 17, 2002.
- Copyright © 2002 by The American Association of Immunologists