Identification and Functional Characterization of a Smad Binding Element (SBE) in the JunB Promoter That Acts as a Transforming Growth Factor-β, Activin, and Bone Morphogenetic Protein-inducible Enhancer*

Smad proteins have been identified as mediators of intracellular signal transduction by members of the transforming growth factor-β (TGF-β) superfamily, which affect cell proliferation, differentiation, as well as pattern formation during early vertebrate development. Following receptor activation, Smads are assembled into heteromeric complexes consisting of a pathway-restricted Smad and the common Smad4 that are subsequently translocated into the nucleus where they are thought to play an important role in gene transcription. Here we report the identification of Smad Binding Elements (SBEs) composed of the sequence CAGACA in the promoter of theJunB gene, an immediate early gene that is potently induced by TGF-β, activin, and bone morphogenetic protein (BMP) 2. TwoJunB SBEs are arranged as an inverted repeat that is transactivated in response to Smad3 and Smad4 co-overexpression and shows inducible binding of a Smad3- and Smad4-containing complex in nuclear extracts from TGF-β-treated cells. Bacterial-expressed Smad proteins bind directly to the SBE. Multimerization of the SBE creates a powerful TGF-β-inducible enhancer that is also responsive to activin and BMPs. The identification of the sequence CAGACA as a direct binding site for Smad proteins will facilitate the identification of regulatory elements in genes that are activated by members of the TGF-β superfamily.

The product of the JunB gene is a member of the AP-1 1 family of transcription factors that activate transcription by binding to TPA response elements (TREs) within the promoter of target genes (1). AP-1 components are immediate early gene products whose expression is rapidly induced by a variety of extracellular stimuli and are encoded by the Fos and Jun families of genes that have been shown to be involved in growth control and differentiation (2). JunB differs in biological properties from its homologs and appears to be a negative regulator of AP-1 function (3). This functional difference is because of a small number of amino acid changes in its DNA binding and dimerization motifs compared with the corresponding c-Jun sequences, as well as to differences in phosphorylation status in response to mitogenic stimulation (4).
The action of JunB as a negative regulator of TRE response elements is consistent with its induction by negative regulators of cell growth including transforming growth factor-␤ (TGF-␤) as well as the structurally and functionally related factors activin and bone morphogenetic protein (BMP) 2/4 (5)(6)(7)(8). In this respect, JunB is a member of the group of genes that are known to be induced in response to TGF-␤ stimulation, which further include the cyclin-dependent inhibitors (CDI) p15 and p21 (9,10) and the plasminogen activator inhibitor (PAI-1) gene (11) that control, in part, cell cycle progression and extracellular matrix remodeling in response to TGF-␤, respectively. However, it is presently unclear whether the induction of these genes by TGF-␤ and related factors involves a direct mechanism.
TGF-␤ family members exert their cellular effects (12,13) by binding to transmembrane receptors that possess serine/threonine kinase activity (14). Upon ligand binding, a heteromeric receptor complex consisting of two type II and two type I receptors is formed. Within the complex, the type I receptor is phosphorylated and activated by the type II receptor constitutively active kinase. Genetic studies in Drosophila melanogaster and Caenorhabditis elegans have recently led to the identification of a conserved family of proteins termed Smads that play an important role in intracellular signal transduction of serine/threonine kinase receptors (15). At present, at least nine family members have been identified in vertebrates. Smads are 40 -62-kDa proteins with N-and C-terminal homology domains (MH1 and MH2) connected by a proline-rich linker. Smad1, Smad5, and presumably MADH6/Smad9 associate with and are phosphorylated after BMP-mediated BMP-RI and ActR-I activation, whereas Smad2 and Smad3 are phosphorylated after TGF-␤R-I and ActR-IB activation. Following phosphorylation, which occurs at a conserved SSXS motif at the extreme C termini, these pathways-restricted Smads form heteromeric complexes with the common mediator SMAD4 and translocate to the nucleus to regulate gene tran-* This work was sponsored by grants from the European Unity (BioMed Program BMH-CT95-0995) and the Dutch Cancer Society (KWF92-84) (to L. J., and W. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AJ004891.
Support for a role of Smads as transcription factors has been obtained from a number of studies. The C-terminal domains of Smad1 and Smad4 have transactivation activity when fused to the Gal4-DNA binding domain in a Gal4-reporter transactivation assay (20,21). Smad2 and Smad4 together with Fast-1, a winged-helix DNA binding protein, associate into activin response factor (ARF) that binds to the activin response element of the Xenopus laevis Mix.2 promoter (22,23). TGF-␤ as well as Smad3 and Smad4 overexpression transactivate the PAI-1 and p3TP-lux promoters (24), which has been attributed to potentiation of AP-1-dependent transcription activation (25). By contrast, Drosophila Mad domain binds a GϩC-rich sequence in the Drosophila vestigial quadrant enhancer (26). However, these investigations have not been conclusive regarding the role of Smads in transcriptional activation, as well as regarding the sequences with which they interact.
In this report, we have investigated the transcriptional regulation of the immediate early gene JunB by TGF-␤. Transient overexpression of Smad3 and Smad4 with various JunB promoter constructs led to the identification of a region in the upstream part of the promoter that is both Smad-transactivated and shows TGF-␤-induced binding of a Smad-containing complex. Further characterization led to the identification of an inverted hexanucleotide repeated sequence binding a TGF-␤-induced DNA binding activity. Multimerization of this sequence created a powerful TGF-␤-inducible enhancer. Implications of these findings for Smad DNA binding and transcriptional activation will be discussed.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-An 8-kilobase EcoRI fragment was isolated from the Balb/c mouse genomic JunB clone 31 which has been described previously (27). All sequences upstream from the initiator ATG were cloned in the pGL3-basic reporter vector (Promega). For cloning purposes, the ATG was converted to ATC by PCR (pJB1). A deletion series was constructed by restriction digestion of the pJB1 plasmid with EcoRI-PinAI (pJB2), SacI (pJB3), SacI-BssHII (pJB4), or SacI-SacII (pJB5) and subsequent self-ligation. A minimal promoter construct (pGL3ti) was made from pGL3-basic by inserting oligonucleotides carrying the adenovirus major late promoter TATA box (AdTu, gatctGGGGCTATAAAAGGGGGTAGGGGgagct, and AdTl, cCCCCTA-CCCCCTTTTATAGCCCCa) and the mouse terminal deoxynucleotidyl transferase gene initiator sequence (Tiu, cGCCCTCATTCTGGAGAC-Ag, and Til, gatccTGTCTCCAGAATGAGGGCgagct) in the BglII site. A deletion series (pJB11-17) was constructed as follows: an Asp718-BseAI (Ϫ3004/Ϫ1561) fragment from pJB3 was inserted into the Asp718-XmaCI sites from pGL3ti (pJB11), a T4 DNA polymerase-blunted Sac-I-BamHI fragment from pJB3 was inserted in the SmaI site of pGL3ti (pJB12), a BamHI-BglII fragment from pJB11 was inserted into the BglII site of pGL3ti (pJB13), pJB12 was restriction-digested with Asp718-PvuII or PvuII-BglII and subsequently blunted and self-ligated (pJB14 and pJB15, respectively), pJB15 was restriction digested with Asp718-MluNI and subsequently blunted and self-ligated (pJB17), and pJB11 was digested with MluNI and BglII and subsequently blunted and self-ligated (pJB16). For fine mapping of the Smad3 and Smad4 response site in the Ϫ2908/Ϫ2611 region, pJB15 was used as template for PCR reactions with the Til oligonucleotide and the upper strand of oligonucleotide D (see "Electrophoretic Mobility Shift Assays"). The PCR product was restriction-digested with SacI and ligated into an Asp718-digested and blunted, SacI-digested pGL3ti vector (pGL3ti-DPv). pGL3ti-PsPv and pGL3ti-AfPv were constructed by digestion of pJB15 with Asp718-PstI or Asp718-AflII, blunting and self-ligation. The internal deletion series was constructed as follows: a PstI/PvuII fragment (Ϫ2762/Ϫ2611) from pJB12 was cloned into the PstI and EcoRV sites of pBluescript SKII-(pSK-PsPv). Next, pJB15 was used as template for PCR reactions with the upper strand of oligonucleotide A and the lower strands of oligonucleotides C, D, and G. The PCR products were cloned into the SmaI site of pSK-PsPv. The resulting plasmids were restriction-digested with XbaI and SalI, and the fusion fragments were inserted into NheI-XhoI-opened pGL3ti. pGL3ti-(SBE) 4 was constructed by inserting two WT oligonucleotides (see "Electrophoretic Mobility Shift Assays") into the XhoI site of pGL3ti. All constucts were verified by sequencing. The nucleotide sequence of the Ϫ2908/Ϫ2611 MluNI/PvuII Smad-responsive region appeared to differ at several positions from a previously deposited JunB genomic sequence (GenBank accession number U20735). The nucleotide sequence of this region as determined by us has been deposited in the EBI Data Bank (accession number AJ004891). Smad expression plasmids were constructed as follows: pSG5-XMAD1 and pSG5-XMAD2 were created by inserting EcoRI fragments from pSP64TEN-DOT1 and -DOT2 into EcoRI-opened pSG5, pSG5-hSmad3f was created by inserting a blunted BamHI-SalI fragment from pRK5-hSmad3f into blunted BamHIopened pSG5, and pSG5-hSmad4 was created by inserting a blunted BamHI-EcoRI fragment from pDPC-wt3 into blunted EcoRI-BamHIopened pSG5. The expression plasmid for Flag-Smad1, Smad2, Myc-Smad3, and Smad4 were described previously (28).
Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared from Mv1Lu cells using a modified Dignam protocol as described previously (30). Oligonucleotides and dephosphorylated restriction fragments were labeled with [␥-32 P]ATP and T4 polynucleotide kinase. Oligonucleotides used in electrophoretic mobility shift assay experiments are shown in Scheme 1. Binding reactions contained 4 g of nuclear extract, 100 mM KCl, 0.2 mM EDTA, 0.5 mM dithiothreitol, 20% glycerol, 20 mM HEPES, pH 7.9, 100 ng of poly(dI-dC) and approximately 10.000 cpm labeled probe and 500-fold molar excess of competitor oligonucleotide where appropriate. Protein binding was allowed to proceed for 30 min at room temperature. Then 20% Ficoll was added to the reactions, and samples were immediately loaded onto 4.5 or 5% polyacrylamide gels containing 0.5ϫ TBE. Smad3 and Smad4 antibodies (DHQ and HPP, respectively) (28) were added undiluted at 0.5 l, after proteins were allowed to bind the probe for 15 min. Samples were then incubated for 15 min before loading onto the gel.
GST fusion protein expression was induced in logarithmically growing cultures of Escherichia coli (TG1) by the addition of isopropyl-D-thiogalactopyranoside to a final concentration of 0.1 mM and growing the bacteria at 30°C for an additional 5 h. Bacteria suspended in 50 ml of cold phosphate-buffered saline were sonicated, mixed with 1% Triton X-100, and centrifuged at 12,000 ϫ g for 10 min. Subsequently, an extract containing GST fusion protein was mixed with 0.25% glutathione-Sepharose 4B beads (Pharmacia) and incubated at 4°C under the constant agitation for 12 h. The beads were washed four times with phosphate-buffered saline containing 1% Triton X-100, and then eluted three times with 50 mM Tris (pH 8.0) including 10 mM glutathione. After the dialysis of eluates with phosphate-buffered saline containing 2 mM dithiothreitol and 0.5 mM phenylmethylsulfonyl fluoride, the sample was stored at Ϫ70°C until used. GST fusion proteins were mixed with the binding buffer consisting of 20% glycerol, 20 mM Hepes (pH 7.9), 30 mM KCl, 4 mM MgCl 2 , 0.1 mM EDTA, 0.8 mM sodium phosphate, 4 mM spermidine, 0.3 g/l poly(dI-dC) (Pharmacia), and 0.25 nM 32 P-labeled probe at the final volume of 20 l. When necessary, 500-fold molar excess of the cold competitor was added to the reaction mixture. The mixture was incubated at 25°C for 1 h and run on a 5% nondenaturing polyacrylamide gel with 0.5ϫ TBE as the running buffer.

RESULTS
Identification of a Smad-responsive Region in the JunB Promoter-Previously, we have shown that the immediate early gene JunB is a direct target for transcription activation in response to activation of signal transduction by TGF-␤ (5). To investigate the mechanism of transactivation of the JunB promoter by TGF-␤, a JunB promoter-luciferase reporter construct was made containing the JunB TATA box and transcription start site, the complete 5Ј-untranslated region as well as approximately 6.4 kilobases of promoter upstream sequences (pJB1). Transient transfection of this construct in NIH3T3, Mv1Lu, and HepG2 cells and treatment of the cells with TGF-␤ did not result in a significant increase in activation of the luciferase reporter construct over uninduced levels after normalization with ␤-galactosidase activities from cotransfected control LacZ expression plasmid (data not shown). Recently, we and others have shown that transiently overexpressed Smad proteins transactivate target genes in a ligand-independent manner (24,28). We therefore cotransfected pJB1 with plasmids expressing individual Smads or combinations of each pathway-restricted Smad and Smad4 into NIH3T3 cells (Fig.  1A). A 3-to 5-fold activation of the reporter construct was observed when Smads 1, 2, and 3 were co-expressed with Smad4 while individual Smads did not significantly activate the reporter. Similar results were obtained using P19 embryonal carcinoma (EC) cells and HepG2 cells. We localized the Smad-responsive region in the JunB promoter by transfecting cells with a series of deletion constructs along with Smad3 and Smad4, which is the strongest activating combination (see Fig.  1A). This analysis showed that the Smad-responsive region is located between nucleotides Ϫ3004 and Ϫ1534; whereas pJB3 has Smad inducibility very similar to pJB1, this response was lost for the JB4 deletion construct (Fig. 1B). A nested set of restriction fragments was derived from this Smad-responsive region and cloned in front of a heterologous minimal promoter fused to the luciferase gene. Cotransfection of these constructs with Smad3 and Smad4 expression plasmids allowed localization of a Smad-responsive region to between nucleotides Ϫ2908 and Ϫ2611 (Fig. 1C). The position of the Smad-responsive element within the Ϫ2908/Ϫ2611 region was determined using a series of progressive 5Ј and internally deleted constructs. Deletion of sequences upstream from nt Ϫ2813 did not affect Smad responsiveness while additional deletion of 26 bp (to nt Ϫ2787) abrogated inducibility. This analysis was complemented by a series of deletions of sequences located between nt Ϫ2908 and Ϫ2762. Exclusion of sequences upstream from nt Ϫ2792 abrogated transactivation by Smad3 and Smad4 (Fig. 1D). This analysis therefore defines the 22-base pair region between Ϫ2813 and Ϫ2792 as that minimally required for transactivation by Smad3 and Smad4. Interestingly, the 22-bp sequence contains a perfect 7-bp inverted repeat (CAGACAGtCTGTCTG).
JunB Promoter Fragments Bind TGF-␤-induced Complex Containing Smad3 and Smad4 -To investigate whether TGF-␤ induces binding of nuclear proteins to the Smad-responsive region, we succesively incubated labeled probes containing nt Ϫ2908 to Ϫ2611, nt Ϫ2908 to Ϫ2788, and nt Ϫ2813 to Ϫ2792 with nuclear extracts from TGF-␤-treated or -untreated Mv1Lu, HaCaT, and NIH3T3 cells cells (Fig. 2). Electrophoretic mobility shift assay experiments showed that extracts from TGF-␤-treated cells contain an induced DNA binding activity that migrates with a lower mobility than that of a constitutively expressed binding entity. These results suggest that TGF-␤ activates the endogenous JunB gene by inducing binding of a nuclear factor to a JunB promoter distal element located between nucleotides Ϫ2813 and Ϫ2792. This fragment correlates with the minimal region required for transactivation by Smad3 and Smad4, suggesting that Smads may form part of nuclear complexes that bind to the 22-bp JunB promoter sequence. To analyze whether Smads are present in nuclear complexes that bind to the 22-bp JunB promoter sequence, antisera directed against Smad3 or Smad4 were added to the binding reactions of the 22-bp JunB probe with nuclear extracts (Fig. 2C). Both Smad3 and Smad4 antisera supershifted the slowest migrating TGF-␤-induced complex. The Smad3 antiserum was more effective than the Smad4 antiserum, which might be due to intrinsic higher affinity of Smad3 antiserum versus Smad4 antiserum. Both antisera produced a supershifted band with extracts from uninduced cells. Possibly, these complexes were formed by stabilization by the antisera of the interaction of contaminating cytoplasmic Smads with the probe. Neither Smad1, 2, or 5 antiserum produced supershifted bands (data not shown). These results indicate that TGF-␤ induces nuclear translocation of activated Smad3 and Smad4 and their subsequent binding to a defined region in the JunB promoter.
Smad Proteins Bind Directly to CAGACA Elements-Having identified Smad3 and Smad4 as components of the TGF-␤induced complexes, we next investigated whether Smad proteins directly bind the Smad-responsive fragment. We pro-SCHEME 1.
duced C-terminal truncated (⌬MH2) and full-length Smads 1-5 as GST fusion proteins and analyzed these for their ability to bind the 120-bp MluNI/AflII (Ϫ2908/Ϫ2788) fragment (Fig.  3A). Smad3⌬MH2 and Smad4⌬MH2 strongly interacted with the probe while binding of C-terminal truncated Smads 1, 2, and 5 to the 120-bp fragment was below the threshold of detection. Binding of full-length Smad4, but not Smad3 fusion protein, was observed, indicating that Smad3 requires a modification or a conformational change to demonstrate its DNA binding potential. These results show that Smad3 and Smad4 bind directly to their target sequence in the 120-bp JunB promoter fragment.
We subsequently tested a series of partially overlapping oligonucleotides (Scheme 1, oligos A, B, C, D, and G) corresponding to the 120-bp fragment for Smad3 and Smad4 binding (Fig. 3, B and C). GST-Smad3⌬MH2 and to a lesser extent GST-Smad4⌬MH2 bound oligonucleotides D and G, whereas oligonucleotide A weakly interacted with GST-Smad3⌬MH2. These results identify the region of overlap between oligos D and G as the major binding site and oligo A as the minor Smad binding site. When comparing the sequence of oligonucleotides A, D, and G, we noted a hexameric CAGACA sequence that was present once in oligonucleotide A and twice in the 7-bp inverted repeat in oligonucleotides D and G. Thus, bacterial-produced Smad3 and Smad4 bind the same 22-bp region that was found to be transactivated by Smad3 and Smad4 and that showed TGF-␤-inducible Smad binding. We will refer to this CAGACA sequence as the Smad binding element (SBE).
The Central GAC Nucleotides in the SBE Are Most Important for Smad3 and Smad4 Binding-To identify bases in the SBE that are important for Smad3 and Smad4 binding, we tested the efficiency of binding of Smad3 and Smad4 to direct repeats of the SBE or mutated versions thereof in which each repeat carried a different substitution of one of the base pairs of the 7 bp of the inverted dimeric SBE (Fig. 4A). The direct repeat and the inverted repeat bound Smad3 and Smad4 equally well (data not shown). The order of binding strength of Smad proteins to the oligonucleotides was as follows: Complementary experiments in which the binding of GST-Smad proteins was challenged by incubation with excess unlabeled wild type or mutant oligonucleotides as competitors yielded results with similar affinity differences between wild type and mutant probes (Fig. 4B). Essentially the same results were obtained for binding of proteins from nuclear extracts, which include Smad3 and Smad4, isolated from TGF-␤-induced cells to the 22-bp inverted repeat-  Fig. 1C) were generated by PCR using oligonucleotide D and a downstream primer or restriction digestion with AflII or PstI. In addition, a series of internal deletion fragments were made by PCR using oligonucleotides A and C, D, and G and fused to the PstI-PvuII (PsPv) fragment. The fragments were ligated in front of the heterologous minimal promoter of the pGL3ti reporter plasmid. The resulting plasmids with or without Smad3 and Smad4 expression plasmids were transfected into P19EC cells. Normalized luciferase activities are shown as the mean Ϯ S.D. of triplicates. The 22-bp region of overlap between oligonucleotides G and D was defined as a Smad3-and Smad4-responsive region. The sequence of this region is shown below the schematic reporter constructs. The arrows denote a 7-bp inverted repeat. containing probe (data not shown). When we tested the Smad binding to a probe containing four directly repeated SBEs, we found that Smad3⌬MH2, Smad4⌬MH2, and full-length Smad4 bound with much higher affinity than two repeats (Fig. 4C). The same results were obtained when the SBEs were present as two indirect repeats (data not shown). In addition, weak binding of full-length Smad3 was now observed. We could specifically compete the Smad3 or Smad4 binding with wild type oligonucleotide, but not with a mutant version in which the G3 and C5 were substituted, and no binding to mutated probe was observed (Fig. 4C). This analysis defines the central nucleotides GAC in the SBE as most important for binding Smad proteins and the TGF-␤-induced complex, while the flanking nucleotides contribute less to the binding and the most 3Ј nucleotide (G7) is not essential.
The SBE Is a TGF-␤ Response Element-To investigate whether the SBE can confer inducibility to TGF-␤, we inserted the SBE in different copy number in front of a minimal promoter and tested these constructs in HepG2 cells. We found that two copies were insufficient for significant induction of luciferase activity, but that with four copies a strong response was obtained (Fig. 5A). Therefore, multimerization of the SBE is sufficient to confer TGF-␤ inducibility to a minimal promoter. When the (SBE) 4 reporter was cotransfected with Smads, we found that Smad3 alone, but not Smad2 or Smad4 alone, had a slight ligand-independent effect. The TGF-␤-induced effect was enhanced by cotransfecting the Smad3 and Smad4 combination, but not with the Smad2 and Smad4 combination. Highest TGF-␤-induced transcriptional response was obtained when all three Smads were cotransfected (Fig. 5A).
Smads Participate in TGF-␤-induced SBE-mediated Transcription-TGF-␤ failed to induce the (SBE) 4 reporter in MDA-MB468 cells that lack the Smad4 gene (31) (Fig. 5B). As the TGF-␤ response was rescued by Smad4 transfection, Smad4 appears to be required for the TGF-␤-induced, SBE-mediated transcriptional response. Further transfection studies showed that Smad3 and Smad4, but not Smad2 and Smad4, cooperated in the SBE-mediated transcriptional response in these cells, which was readily observed in the absence of ligand. In fact, cotransfection of Smad3 and Smad4 resulted in a level of reporter activation over which no dramatic TGF-␤-dependent respone was observed. Cotransfection of the reporter plasmid into HepG2 cells along with increasing amounts of a Smad7 expressing plasmid showed a dose-dependent decrease in TGF-␤-induced SBE-mediated transcriptional response (data not shown). Taken together, the results indicate that Smad3 and Smad4 are directly involved in the T␤R-I-mediated activation of the JunB promoter-derived SBE.
SBE Is a Response Element for Other Members of the TGF-␤ Family-Activin and BMP2 have been shown to induce JunB mRNA expression (7,8). To test whether the SBE can confer inducibility to activin and BMP, we transfected HepG2 cells with the TGF-␤-responsive (SBE) 4 construct and treated the transfected cells with either OP-1 (also termed BMP-7), activin, or TGF-␤ (Fig. 5C). All three ligands activate the SBE reporter albeit with different efficiencies. To corroborate these findings, we transfected Smad4-negative MDA-MB468 cells with the CAGACA reporter plasmid with or without the Smad4 expression plasmid and tested the transfected cells for responsiveness to TGF-␤, activin, OP-1, BMP2, and GDF5 (Fig. 5D). In the absence of Smad4, the reporter construct was virtually unresponsive to any of the ligands. Cotransfection of Smad4 resulted in a slight activation of the reporter. However, treatment of the Smad4 cotransfected cells with the various members of the TGF-␤ superfamily showed that OP-1 and TGF-␤ strongly activated the (SBE) 4 construct, whereas treatment with activin was without effect, which may indicate that our MDA-MB468 cells lack functional activin receptors. Furthermore, BMP2 and GDF5, which bind the BMP type IB receptor (BMPR-IB) (32) also strongly activated the (SBE) 4 . In agreement with these findings, constitutively active versions of activin type I receptor (ActR-I), ActR-IB, BMPR-IA, and TGF-␤ type I receptor induced (SBE) 4 mediated transcription in HepG2 cells (data not shown). Taken together these results indicate that the (SBE) 4 can be activated by different members of the TGF-␤ superfamily. DISCUSSION TGF-␤ is the prototype of a family of signaling molecules that exhibit pleiotropic effects on cell proliferation, differentiation, as well as on pattern formation during early vertebrate development (33). The recent discovery of Smad proteins as downstream effectors of signaling by activated receptors of members of the TGF-␤ superfamily has opened new ways for investigating regulation of target gene expression. Taking advantage of the capacity of Smad proteins to act as ligand-independent activators of gene expression when transiently overexpressed (24, 28), we have identified short sequences in the mouse JunB gene promoter termed Smad binding elements (SBEs) that mediate responsiveness to activation of intracellular signal transduction by several members of the TGF-␤ superfamily. We have shown that two SBEs in the JunB promoter form an inverted hexameric CAGACA repeat that readily binds a Smad3-and Smad4-containing complex from TGF-␤-treated Mv1Lu cells. Bacterial-expressed full-length Smad4 and Cterminal truncated Smad3 or Smad4 bind to the JunB SBE repeat, demonstrating that the identified CAGACA sequence is a direct binding site for Smads. Binding of Smad proteins is independent of the relative orientation of the SBEs in a repeat, but multimerization of SBEs strongly increases the affinity for Smad proteins, indicating that cooperative binding is required for Smad function. This is demonstrated by the fact that activation of SBE-containing reporter constructs by TGF-␤ or overexpressed Smad proteins is only observed when four CAGACA elements are present.
The JunB gene is a target for BMP2, activin, and TGF-␤signaling (5)(6)(7)(8). Ligand-mediated activation of the endogenous promoter can be mimicked by cotransfection of a reporter construct with the pathway-restricted Smad1, Smad2, and Smad3 along with Smad4. The CAGACA repeat isolated from the JunB gene, as present in the pGL3ti-(SBE) 4 reporter construct, is also activated by TGF-␤, activin, BMP2, OP-1/BMP7, and GDF5, implicating that these ligands activate the JunB promoter through the same response element. Members of the BMP-subfamily have been reported to signal through Smad1 and Smad5, while activin and TGF-␤-signaling is mediated by Smad2 and Smad3 (15). Remarkably, the SBE repeat is only efficiently bound by bacterially produced Smad3 and Smad4. Possibly, Smads 1, 2, and 5 require additional proteins for high affinity binding to the SBE. Alternatively, and in contrast to Smad3, these Smad proteins may associate with their target sequence only through Smad4. Treatment of cells with TGF-␤ induces the phosphorylation of both Smad2 and Smad3, which share a high degree of homology (24). However, we could not detect Smad2 protein in the TGF-␤-induced complex formed with the 22-bp probe. The major difference between Smad2 and Smad3 resides in the DNA-binding MH1 domain. Possibly, the insertion in this domain of Smad2 alters its binding characteristics.
The response element we have isolated from the JunB promoter shows no homology with Mad binding site in the Drosophila decapentaplegic-responsive vestigial quadrant enhancer has a GC-rich core and shows no homology with the Smad binding sequence as derived from the JunB promoter (26). Interestingly, like Smad3, Mad binds its recognition sequence only as a C-terminal truncated protein. These properties may be related to a proposed working mechanism for receptor-activated Smads, in which Smads become signalingcompetent only when phosphorylation relieves the inhibitory action of their C-terminal MH2 domain. The sequence of the JunB SBE binding sequence shows no homology to Sp1-binding sequences that previously have been implicated in TGF-␤-induced transactivation of the p15 and p21 promoters (9,10). Smad2 and Smad4 interact with FAST-1 to form ARF, which binds through FAST-1 to the ARE in the X. laevis Mix.2 promoter (23). The JunB promoter nor the pGL3ti-(SBE) 4 reporter are transactivated by FAST-1 in HepG2 cells, in contrast to a construct containing a multimerized ARE (19). 2 The ARE has been reported to contain two 6-bp sequences that are both required for FAST-1 binding and ARF formation. These sequences do not resemble the JunB SBE. However, one of the 6-bp sequences overlaps with a perfect SBE sequence. It will be of interest to investigate whether this SBE sequence binds either Smad2 or Smad4 present in ARF. Recently, a Smad 2 L. J. C. Jonk and W. Kruijer, unpublished observations. FIG. 3. Bacterially expressed Smad proteins directly bind to the Smad-responsive element. A, binding of Smad proteins to the 120-bp MluNI-AflII region. Smads 1-5 were expressed as full-length or C-terminal truncated (⌬MH2) GST-fusion proteins in E. coli. The purified proteins were incubated with the labeled probe and complexes were resolved on a 5% nondenaturing polyacrylamide gel. Only full-length Smad4 and Smad3⌬MH2 and Smad4⌬MH2 interacted with the probe. B, mapping of the binding site for bacterially produced Smad3 and Smad4. C-terminal truncated Smad3 or Smad4 proteins were incubated with end-labeled double-stranded oligonucleotides corresponding to A, B, C, D, and G (see Fig. 1D), and complexes were resolved on a nondenaturing 5% polyacrylamide gel. Strong binding of Smad3⌬MH2 and Smad4⌬MH2 was observed only for oligonucleotides D and G.
binding site was identified in the TGF-␤-responsive promoter of the p3TP-Lux construct (25). This site contains a monomeric CAGACA sequence identical to that of the JunB SBE and overlaps with an AP-1 site. The CAGACA sequence appeared to be dispensable for TGF-␤ induction because mutation of this sequence did not affect TGF-␤ responsiveness. However, the mutated construct used in this experiment still contained a CgGACA sequence that may be a binding site for Smad proteins. By contrast, mutation of the AP-1 site completely abrogated TGF-␤ induction. Possibly, in the context of this promoter, Smad proteins cooperate with AP-1 to mediate TGF-␤ induction. In this respect, it is noteworthy that the TGF-␤inducible PAI-1 promoter contains three CAGACA elements that were able to mediate TGF-␤, but not BMP-responsiveness, 3 of which one element is located adjacent to an AP-1 site. Likewise, the human ␣2(I) collagen promoter also harbors an AP-1 site in close proximity to a monomeric CAGACA sequence probe. Binding of truncated Smad proteins and full-length Smad4 is efficiently competed by wild type (WT) but not mutant (MT) SBEcontaining oligonucleotide. Full-length Smad3 displays very weak binding. The labeled mutant probe does not bind GST-Smad protein.  4 plasmid with or without a Smad4 expression plasmid. Following transfection, cells were incubated for 16 h with or without 10 ng/ml TGF-␤1, 20 ng/ml activin, 100 ng/ml OP-1, 100 ng/ml BMP2, or 200 ng/ml GDF5. Normalized luciferease activities are expressed as the mean Ϯ S.D. of triplicates. (34). Finally, Smad proteins may regulate gene transcription through binding to AP-1 or other transcription factors without interacting with DNA in a manner analogous to AP-1/steroid hormone receptor associations. The findings may hint to a mechanism for Smad-mediated gene activation in which receptor-activated Smad proteins cooperate with ubiquitous transcription factors. Nuclear extracts from untreated and TGF-␤treated cells contain a constitutive binding activity that complexes with the 22-bp inverted CAGACA repeat-containing probe. The constitutive complex cannot be supershifted with anti-Smad antibodies. Although we cannot rule out the possibility that the constitutive activity corresponds with a Smadrelated protein, in untreated cells the SBE may contain a novel factor that is pre-bound to DNA to which nuclear translocated Smads bind to form a transactivating complex. We are presently studying the JunB Smad-responsive region by genomic footprinting to resolve this issue. This complex organization may explain our inability to show TGF-␤-induction of our transiently and stably transfected JunB reporter constructs. Imitation of the exact chromosomal context of the endogenous JunB gene may be required for investigation of regulation of this gene by members of the TGF-␤ superfamily. Nevertheless, the identification of the SBE as a binding sequence for Smad proteins now opens the way to investigate the binding determinants of Smad proteins by site-directed mutagenesis and elucidation of the crystal structure of the Smad/DNA complex. Furthermore, the interactions of Smad proteins with other proteins jointly forming a transcriptionally active complex in response to members of the TGF-␤-superfamily can now be determined.