Regulation of Smad7 Promoter by Direct Association with Smad3 and Smad4*

Smad7 is a regulatory Smad protein that is able to antagonize signal transduction by transforming growth factor-β (TGF-β) and activin receptors. To characterize the regulation of Smad7 at the transcriptional level, we isolated the promoter region of the mouse Smad7 gene. When the Smad7 promoter luciferase reporter gene (−408 and +112 bp) was expressed in human hepatoma (HepG2) cells, its transcriptional activity was increased following TGF-β or activin treatment. In addition, this region of the Smad7 promoter was stimulated by ectopic expression of Smad3 as well as constitutively active TGF-β and activin receptors, indicating that Smad7 transcription was modulated by the signaling downstream those two receptors. A gel mobility shift assay indicated that a DNA fragment spanning −408 to −126 base pairs (bp) was able to directly bind purified Smad4. Furthermore, a consensus Smad3-Smad4 binding element (SBE) was discovered in this region of the promoter with a palindromic sequence of GTCTAGAC. A 33-bp Smad7 promoter fragment containing this SBE was able to bind Smad3 and Smad4. In human embryonic kidney 293 cells, the expression of constitutively active TGF-β type I receptor was able to induce the formation of a Smad3- and Smad4-containing nuclear protein complex that bound the SBE. In HepG2 cells, TGF-β1 treatment could induce the formation of an endogenous SBE-binding complex. Taken together, these data provided the first evidence that Smad7 transcription is regulated by TGF-β and activin signaling through direct binding of Smad3 and Smad4 to the Smad7 promoter.

Smad7 is a regulatory Smad protein that is able to antagonize signal transduction by transforming growth factor-␤ (TGF-␤) and activin receptors. To characterize the regulation of Smad7 at the transcriptional level, we isolated the promoter region of the mouse Smad7 gene. When the Smad7 promoter luciferase reporter gene (؊408 and ؉112 bp) was expressed in human hepatoma (HepG2) cells, its transcriptional activity was increased following TGF-␤ or activin treatment. In addition, this region of the Smad7 promoter was stimulated by ectopic expression of Smad3 as well as constitutively active TGF-␤ and activin receptors, indicating that Smad7 transcription was modulated by the signaling downstream those two receptors. A gel mobility shift assay indicated that a DNA fragment spanning ؊408 to ؊126 base pairs (bp) was able to directly bind purified Smad4. Furthermore, a consensus Smad3-Smad4 binding element (SBE) was discovered in this region of the promoter with a palindromic sequence of GTCTAGAC. A 33-bp Smad7 promoter fragment containing this SBE was able to bind Smad3 and Smad4. In human embryonic kidney 293 cells, the expression of constitutively active TGF-␤ type I receptor was able to induce the formation of a Smad3-and Smad4-containing nuclear protein complex that bound the SBE. In HepG2 cells, TGF-␤1 treatment could induce the formation of an endogenous SBE-binding complex. Taken together, these data provided the first evidence that Smad7 transcription is regulated by TGF-␤ and activin signaling through direct binding of Smad3 and Smad4 to the Smad7 promoter.
Smad proteins are a group of recently identified molecules that function as intracellular signaling mediators and modulators of transforming growth factor-␤ (TGF-␤) 1 family members (1,2). Functional characterization of Smad proteins has allowed their subdivision into three subfamilies: pathway-specific, common mediator, and inhibitory Smads. Pathway-specific Smads are activated by the type I receptor serine kinases through phosphorylation at the C-terminal end in a ligand-and type II receptor-dependent manner (3,4). This subfamily includes Smads 1, 2, 3, 5, and 8 (2). Smad1 and -5 mediate signaling by bone morphogenetic proteins (BMP) 2 and 4 (5-8), Smad2 and -3 mediate signaling by TGF-␤ and activins (9 -12), and Smad8 mediates signaling by the receptor serine kinase ALK-2 (13). The second Smad subfamily is represented by Smad4 (14), which serves as a common signaling mediator. Activation of the pathway-specific Smads by their individual receptors induces an association of these Smads with Smad4, which is critical for the proper downstream signaling (15). Smad6 and Smad7 comprise the third Smad subfamily and have been reported to function as negative regulators of receptor serine kinases mediating TGF-␤ and BMP responses (16 -18). Smad7 has been shown to inhibit signal transduction by the TGF-␤ and activin receptors (16,17,19), whereas Smad6 was reported to inhibit BMP signaling (18). A Xenopus Smad7 homologue was also reported recently to antagonize BMP and activin signaling in the frog embryo (20,21).
Smad proteins, when translocated into the nucleus, function as transcriptional regulators that control the expression of target genes (2,22,23). Smad exerts its transcriptional regulatory activity by interacting with either a specific transcription factor or a specific DNA element. In Xenopus, Smad2 and Smad4 participate in the activin-mediated transcriptional induction of the Mix.2 promoter through an interaction with a specific DNA-binding transcription factor, forkhead activin signal transducer-1 (FAST-1), a member of the winged-helix forkhead transcription factor family (24,25). FAST-1 has two functional domains that mediate DNA binding and Smad association respectively. The DNA binding motif of FAST-1 mediates the interaction of FAST-1 with an activin-responsive element on the Xenopus Mix.2 promoter, and the Smad interaction domain is involved in the association with the Smad2-Smad4 complex (25). Smad proteins have been found to interact functionally with other transcription factors including human FAST-1 and mouse FAST-2 (26 -28), AP1 (29), Sp1 (30), and TFE3 (31). In addition to the interaction of Smad with other transcription factors, recent studies have suggested that Smad can directly bind DNA. In Drosophila, Mad protein is able to directly bind a GC-rich region of various enhancers (32). A PCR-based screening with random sequences has led to the discovery of specific binding of Smad3 and Smad4 with a palindromic DNA sequence (33). Smad3 and Smad4 were also found to bind a CAGA motif in the promoter of plasminogen activator inhibitor-1 (34). Once Smad protein associates with either a transcription factor or a DNA element, its C-terminal MH2 domain may exert a transactivation function (6). This function of the MH2 domain was supported by the recent discovery that Smad may interact with the general transcription coactivators CBP (CREB-binding protein) and p300 (35)(36)(37)(38). CBP/p300 may bridge the general transcription machinery and Smad proteins or the Smad-associated transcription factors, e.g, FAST-1 or FAST-2, and enable the transcriptional regulation of target genes.
Recent studies have indicated that the inhibitory Smad proteins may play a role in a negative feedback loop that modulates the signaling by TGF-␤, activin, and BMP. Treatment of Mv1Lu mink lung cells and HaCaT keratinocytes with TGF-␤ led to a transient increase of the Smad7 mRNA steady state level (17). In mouse B cell hybridoma HS-72 cells, activin is able to induce an increase of the steady state level of Smad7 mRNA (19). In addition, treatment with BMP2 or BMP7 induces an increase of the Smad6 mRNA level in a number of mouse cell lines (39). Those studies have suggested that the inhibitory Smad proteins could be up-regulated by signaling of the TGF-␤ superfamily; such a regulation might be involved in desensitization of the cells to the continued exposure of ligand. To understand the mechanism underlying the regulation of the inhibitory Smad message, we isolated a mouse Smad7 promoter and characterized the regulation of the promoter by TGF-␤ and activin signaling at the transcriptional level.

MATERIALS AND METHODS
Cell Culture and Cell Transfection-Human embryonic kidney 293 (HEK293) cells and human hepatoma (HepG2) cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum supplemented with penicillin and streptomycin. Transient cell transfection was performed by a calcium phosphate method for HEK293 cells (12) and a DEAE-dextran method for HepG2 cells (40).
Cloning of the Mouse Smad7 Promoter-A DNA fragment corresponding to about 500 bp of the 5Ј-end of a rat Smad7 cDNA was used to screen a mouse 129SvJ genomic library (Stratagene) under high stringency conditions. Ten positive clones were isolated from 5 ϫ 10 5 phages. The inserts of these clones were released by EcoRI digestion and subcloned into KSϩ pBluescript (Stratagene). Analysis by restriction enzyme digestion and partial sequencing of each of those clones revealed that 6 of them have DNA sequences 5Ј to the Smad7 cDNA probe. Both strands of a 520-bp fragment (released by KpnI and XhoI digestion) next to the 5Ј-end of the cDNA probe were completely sequenced (Sequenase version 2, United States Biochemical).
Promoter Assay-The clone that contained the largest 5Ј sequence of the Smad7 gene was 4.3 kb. The whole 4.3-kb fragment contained multiple KpnI sites at the positions Ϫ4.2, Ϫ2.2, and Ϫ408 bp and a single XhoI site at ϩ112 bp relative to the major transcription initiation site. Different lengths of the Smad7 promoter generated by partial digestion by KpnI and complete digestion by XhoI were subcloned into the pGL2-basic luciferase plasmid (Promega). The Ϫ408 to Ϫ124-bp Smad7 promoter/luciferase construct was generated by digestion of the Smad7 genomic clone with KpnI and StuI followed by subcloning into the same luciferase plasmid. The Ϫ283 to Ϫ124-bp promoter construct was generated by XbaI and StuI digestion, and the released fragment was subcloned into KSϩ pBluescript and then cloned into the luciferase plasmid. The Ϫ283 to Ϫ124-bp construct would change the putative Smad binding motif GTCTAGAC of the Smad7 promoter to CTCTA-GAC. For promoter assays, approximately 5 ϫ 10 4 cells/well in 6-well plates were transfected with different combinations of plasmid DNA. PCMV-␤-galactosidase was co-transfected to serve as an internal control for transfection efficiency. The cells were harvested at 48 h after transfection by lysis with 400 l of Nonidet P-40 lysis buffer (15). In TGF-␤-or activin-treated groups, cells were treated with TGF-␤1 (1 ng/ml) or activin A (10 ng/ml) for 12 h before harvest. Twenty l of the lysate was used for the ␤-galactosidase assay as reported previously (12), and 10 l of the lysate was used for the luciferase assay using a Promega luciferase assay kit. The samples were counted for 10 s in a FB12 luminometer (Zylux), and the data were represented as the relative light unit/second.
Primer Extension Assay-An oligonucleotide (5Ј-GCTCGAGTCCT-TCTGCCGCCG-3Ј) corresponding to the very 5Ј-end of the known mouse Smad7 cDNA sequence was labeled at its 5Ј-end by T4 polynucleotide kinase (Promega) in the presence of [␥ 32 P]ATP. The labeled probe was annealed with total RNA (2-4 g) isolated from mouse aorta, mouse whole brain, and C2C12 mouse myoblast cells. The extension reaction was performed in the presence of actinomycin D (20 ng/ml), the four dNTPs (0.5 mM each), 4 mM dithiothreitol, and 20 units of Superscript reverse transcriptase (Life Technologies, Inc.) at 50°C for 30 min. The reaction was then extracted with phenol, precipitated by isopropanol, and loaded on a 6% denaturing polyacrylamide gel. An unrelated plasmid was sequenced and used as a marker to determine the relative position of the primer extension product. Electrophoretic Mobility Shift Assay-A 282-bp fragment (Ϫ408 to Ϫ124 bp) of the Smad7 promoter was released by KpnI and StuI digestion and subcloned into KSϩ pBluescript. This insert was then cut from pBluescript by KpnI and BamHI digestion and labeled with 32 P by a fill-in reaction with the Klenow fragment of DNA polymerase I (Promega). A 33-bp probe containing the putative Smad binding element was generated by annealing two oligonucleotides (5Ј-TCGACAGGGTG-TCTAGACGGCCACG-3Ј and 5Ј-TCGACGTGGCCGTCTAGACACCCT-G-3Ј). This reaction created a 5Ј overhang, which was labeled with 32 P using a fill-in reaction as described above. The probes (5 ϫ 10 4 cpm) were incubated with about 0.2 g of GST or GST-fusion proteins or 2 g of nuclear extracts in a buffer containing a final concentration of 4% glycerol, 10 mM Tris (pH 7.5), 1 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM dithiothreitol, 50 mM NaCl, and 0.1 g/l poly(dI-dC). The nuclear extracts were prepared as described previously by others (41). The reaction was incubated at room temperature for 1 h and then separated by 4% nondenaturing polyacrylamide gel electrophoresis in 0.5ϫ TBE and detected by autoradiography.
Plasmid Construction and GST Fusion Proteins-The wild type Smad2, Smad3, and Smad4 and the constitutively active activin type I receptor have been described previously (12,13). The Smad3 (G/S) mutant was generated by a PCR strategy that specifically changed glycine 379 to serine. The C-terminal truncation of Smad3 was generated by removing the region that spans a HpaI site and the end the cDNA clone. These mutations were confirmed by DNA sequencing. All of the GST fusion proteins were generated by in-frame fusion of the full-length Smad cDNA with pGEX-4T2 (Amersham Pharmacia Biotech). The constructs were transformed into Escherichia coli BL21 strain (Amersham Pharmacia Biotech), and the GST fusion proteins were purified according to the manufacturer's protocol.

Isolation of a Mouse Smad7
Promoter-Smad7 is a newly found regulatory protein that is able to antagonize TGF-␤ and activin signaling (1). Smad7 has been implicated in a negative feedback loop in which TGF-␤ and activin treatment in different cells were able to increase the expression of Smad7 (17,19). To determine whether the regulation of Smad7 expression occurs at the transcriptional level, we isolated the 5Ј region of the mouse Smad7 gene. A 500-bp probe corresponding to the 5Ј-end of rat Smad7 cDNA was used to screen a mouse genomic library at high stringency. Six independent clones that contain sequences 5Ј to the Smad7 cDNA were isolated. Genomic mapping analysis using restriction enzyme digestion and partial sequencing of these clones confirmed that we obtained as much as 4.3 kb of the 5Ј sequence of the Smad7 gene (data not shown). The 512-bp fragment adjacent to the Smad7 cDNA sequence was subjected to full sequencing (Fig. 1A). To identify the putative transcription initiation site, we used a primer extension assay with the total RNA isolated from mouse aorta, mouse whole brain, or the mouse C2C12 myoblast cells. In all of the RNA samples, a major extension product was detected 113 bp upstream of the extension primer ( Fig. 1B) with two minor sites located at 241 and 302 bp, respectively (Fig. 1B). For clarity of data presentation, this major initiation site was designated the ϩ1 position. Sequence information suggests that there was no TATA box in the promoter region. However, the promoter region has multiple Sp1 sites, which are common to most TATA-less promoters (42). In addition, there appears to be a transcription factor IId (TFIID) binding site about 200 bp away from the major initiation site, indicating that the binding of TFIID at this site might contribute to the initiation of Smad7 transcription.
Smad7 Promoter Is Stimulated by TGF-␤ and Activin Signaling-We next determined whether this putative Smad7 promoter was regulated by TGF-␤ and activin signaling. Different lengths of Smad7 promoter sequence were fused with a basic luciferase reporter that did not contain any promoter sequence or TATA box. These reporter constructs spanned from Ϫ4.2 kb, Ϫ2.2kb, or Ϫ408 to ϩ112 bp relative to the major transcription initiation site. When these three luciferase fusion plasmids were transfected in HEK293 cells, all of them gave rise to very high luciferase activity, over 100-fold higher than the value from the cells transfected with the parental luciferase vector (data not shown). These data suggested that the Smad7 genomic clone that we isolated, especially the Ϫ408 to ϩ112-bp region, did contain an endogenous transcription initiation site controlled by the basic transcriptional machinery. In addition, we have made another luciferase fusion plasmid that contains the sequence from the Ϫ2.2 kb to Ϫ408-bp region, and when transfected in HEK293 cells, this construct gave no appreciable promoter activity (data not shown), further indicating that the transcription of the Smad7 gene starts from a region within the Ϫ408 to ϩ112-bp area.
We next transfected these luciferase constructs into human HepG2 cells that contain functional TGF-␤ and activin receptors (30,43). Interestingly, treatment of these cells with activin A or TGF-␤1 was able to stimulate the activity of these pro-moter constructs ( Fig. 2A). Activin or TGF-␤ treatment was able to induce the activity of all three promoter constructs 5-7-fold. Both of the longer constructs had a slightly higher basal activity when compared with the Ϫ408 to ϩ112-bp construct. However, the fold inductions by either activin or TGF-␤ treatment were very similar among the three promoter constructs. These data, therefore, provided the first piece of evidence that the Smad7 promoter could be regulated by TGF-␤ and activin at the transcriptional level. Furthermore, they suggested that the putative regulatory element(s) that mediates the TGF-␤ and activin effect is localized within the Ϫ408 to ϩ112-bp region.
To further characterize the regulation of the Smad7 promoter by TGF-␤ and activin signaling, we analyzed the ability of different Smad proteins and serine receptor kinases of the TGF-␤ superfamily to transactivate the Smad7 promoter (Fig.  2B). Transfection of Smad3, but not Smad1, Smad2, or Smad4 alone, into HepG2 cells was able to stimulate the promoter activity of the Ϫ408 to ϩ112-bp construct. This is consistent with previous findings that Smad3, when overexpressed, is able by itself to up-regulate target genes downstream of TGF-␤ and activin signaling (12). In addition, Smad3 appeared to synergize with Smad4 to stimulate the Smad7 promoter, as co-expression of Smad3 and Smad4 was able to strongly increase the promoter activity (Fig. 2B). As expected, the constitutively active activin type I receptor (ALK-4) and TGF-␤ type I receptor (ALK-5), but not the type I receptor for BMP2/4 FIG. 1. Smad7 promoter sequence  and primer extension assay. A, the Smad7 promoter sequence spanning Ϫ408 to ϩ112 bp is shown here with the putatively transcribed region in uppercase letters. The putative binding sites for transcription factors were determined through the internet by the TESS program (provided by the Computational Biology and Informatics Laboratory at the University of Pennsylvania Schools of Medicine Engineering and Applied Science). The putative Smad binding element is marked in bold letters, and the position of the primer used for the primer extension assay is marked at the end of the sequence. B, results of primer extension assay. Three RNA samples were used in this assay and labeled at the top of the gel. The relative positions of the major and minor transcription initiation sites determined by a DNA sequencing reaction (not shown) loaded next to the three lanes is marked on the right.
(ALK-3) were able to significantly stimulate the Smad7 promoter, although TGF-␤ type I receptor appeared to be stronger than ALK-4 in the transactivation. Interestingly, activation of the Smad7 promoter by ALK-4 and ALK-5 could be antagonized by co-expression of Smad7 (Fig. 2B). In addition, we used dominant negative Smad3 to further determine whether Smad proteins were involved in the activation of the Smad7 promoter by TGF-␤ signaling. Both the Smad3 G/S point mutation and the C-terminal deletion constructs were transfected into HepG2 cells together with the Ϫ408 to ϩ112-bp Smad7 promoter construct (Fig. 2C). The G/S mutation was originally discovered in Drosophila Mad in which it led to a compromised development characteristic of a defective decapentaplegic (dpp) pathway (44), indicating that this mutation is able to disrupt the signaling by TGF-␤ family members. The C terminus or MH2 domain of Smads has been implicated in the transactivating activity of these proteins (6). We found that both G/S point mutation and C-terminal deletion of Smad3 was able to significantly ablate the TGF-␤-mediated activation of Smad7 promoter. Taken together, these data would support a model in which Smad7 is involved in a mutual feedback regulation in TGF-␤ and activin signaling. Activation of TGF-␤ or activin signaling by ligand binding may initiate Smad7 transcription, which may desensitize the further signaling by TGF-␤ or activin. Meanwhile, the shut-off of this signaling by up-regulated Smad7 also blocks its own transcription, making the cells available to respond to new stimuli after a certain delay that would be dependent on the turnover rate of Smad7.
Binding of Smad3 and Smad4 to Smad7 Promoter-Transcriptional regulation by TGF-␤ or activin signaling is achieved mainly by two pathways. One of them is the interaction of Smad proteins with other transcription factors that bind specific sequences of TGF-␤/activin-responsive promoters. The classical example of this category is the FAST-2-mediated transcriptional regulation of the Mix.2 promoter in which Smad2 and Smad4 need to complex with FAST-2 in order to mediate TGF-␤ or activin response (24). The second mode of transcriptional regulation by TGF-␤ or activin is through direct binding of Smad3 and Smad4 with specific DNA sequences. This pathway is exemplified by the finding that these two Smad proteins are able to associate with the promoter of the plasminogen activator inhibitor-1 (PAI-1) gene (34). To explore the possible mechanism underlying the TGF-␤-and activin-mediated transcriptional regulation of the Smad7 promoter, we used a gel mobility shift assay to determine whether Smad proteins directly bind Smad7 promoter. A 282-bp DNA fragment that spans from Ϫ408 to Ϫ126 bp was labeled with 32 P and used in a gel shift assay with Smad2, Smad3, and Smad4 GST fusion proteins (Fig. 3A). Compared with GST alone (Fig. 3A, lane 1), the full-length Smad2 protein could not cause any appreciable shift of the probe (lane 2). The full-length Smad3 protein led to a slightly detectable shift of the probe (lane 3). Furthermore, this 282-bp probe was significantly shifted by the full-length Smad4 fusion protein (lane 7), and this shift could be competed off by the excess of cold 282-fragment (lane 8), supporting the specificity of the binding of Smad4 to the sequence.
Our gel shift assay with the 282-bp probe provided the clue that Smad3 and Smad4 may directly bind the Smad7 promoter and mediate the transcriptional response by TGF-␤ and activin signaling. Careful examination of the Smad7 promoter sequence has led our attention to a putative Smad binding element inside the Ϫ408 to ϩ112-bp region, in which there is indeed a consensus Smad binding sequence previously identified through a PCR-based oligonucleotide screening with the MH1 domain of Smad3 and Smad4 (33). It has been found that both Smad3 and Smad4 could preferentially bind an octamer sequence GTCTAGXC (X stands for any one of the four nucleotides). The Smad7 promoter contains a palindromic sequence of GTCTAGAC in the area spanning Ϫ285 to Ϫ278 bp (Fig. 1A).
To determine whether this putative Smad binding element (SBE) confers the binding of Smad7 promoter to Smad3 and Smad4, we used a synthetic oligonucleotide probe that covered this consensus binding element. As shown in Fig. 3B, this 33-bp probe was not able to bind GST or the Smad2 fusion protein (lanes 1 and 2). However, Smad3 was able to associate with this probe (lane 4), and this binding could be competed off by increasing concentrations of the cold oligonucleotide (lanes 5-7). Furthermore, the binding of Smad3 with the 33-bp probe could be supershifted by an anti-Myc antibody that recognizes the Myc epitope tag of Smad3 (lanes 8-10). In addition to Smad3 binding, the probe could strongly bind the Smad4 (lane 11) that was competed off by the cold probe (lanes 12-14), and it appeared that Smad4 might bind this sequence as a monomer, dimer, or trimer. Taken together, these data strongly suggested that Smad3 and Smad4 were able to directly associate with the consensus Smad binding motif of the Smad7 promoter. To further determine whether the binding of Smad3 or Smad4 with this consensus sequence is involved in the binding of these FIG. 2. Regulation of Smad7 promoter activity by TGF-␤ and activin signaling. A, TGF-␤ and activin stimulation of the Smad7 promoter. Different lengths of the putative Smad7 promoter DNA were fused to a luciferase reporter and transfected into HepG2 cells (0.25 g of promoter construct/well in 6-well plate cotransfected with pCMV-␤galactosidase). The transfected cells were treated with TGF-␤ (1 ng/ml) or activin A (10 ng/ml) for 12 h, and the whole cell lysate was used in luciferase and ␤-galactosidase assays. The fold change of luciferase activity normalized by ␤-galactosidase activity was compared with the values of cells transfected with the shortest promoter construct and without activin treatment (set to 1) and shown as the mean Ϯ S.D. B, regulation of Smad7 promoter by different Smad proteins and TGF-␤ family receptors. HepG2 cells were transfected with the Smad7 promoter Ϫ408 to ϩ112-bp luciferase construct (0.25 g); pCMV-␤-galactosidase (0.25 g); Smad1, Smad2, Smad3, or Smad4 (2 g each); constitutively active ALK-3, ALK-4, or ALK-5 (2 g each); and Smad7 (1 g), as indicated. The activity of the Smad3 and Smad4 co-expression group was much higher than other groups; the relevant data were shown in truncated format with the actual fold induction value indicated at the top of the bar. C, effect of dominant negative Smad3. HepG2 cells were transfected with the Smad7 promoter Ϫ408 to ϩ112-bp luciferase construct (0.25 g) or with Smad3 with G/S point mutation or C-terminal deletion (2 g). The transfected cells were treated with TGF-␤1 (1 ng/ml) for 12 h before luciferase assay.
Smad proteins with the longer Smad7 promoter (i.e. the 282-bp probe), we examined the ability of the 33-bp cold probe to compete with the binding of Smad proteins with the 282-bp fragment. Inclusion of this 33-bp cold oligonucleotide was able to completely compete off the binding of Smad4 with the 282-bp probe (Fig. 3A, lane 9), further indicating that the binding of these Smad proteins to the Smad7 promoter is conferred by the SBE.
We next examined whether an intact SBE was required for the responsiveness of the Smad7 promoter to activin and TGF-␤ treatment. We used two Smad7 promoter constructs, Ϫ408 to Ϫ124 bp and Ϫ283 to Ϫ124 bp. The Ϫ283 to Ϫ124-bp construct had a partial truncation in the 5Ј-end of the SBE, and this truncation changed the putative SBE of the Smad7 promoter from GTCTAGAC to CTCTAGAC (i.e. G to C at the first position). When the Ϫ408 to Ϫ124-bp construct containing the intact SBE sequence was transfected into HepG2 cells, it was stimulated by both activin and TGF-␤ treatment (Fig. 4). However, the Ϫ283 to Ϫ124-bp construct was no longer responsive to activin or TGF-␤, further indicating that this consensus Smad binding motif in the Smad7 promoter was involved in activin-and TGF-␤-mediated transcriptional regulation.
TGF-␤ Receptor Activation Induces a Smad3-Smad4 Complex That Binds Smad7 SBE-Ligand binding of TGF-␤ or activin to membrane receptors is followed by the phosphorylation of Smad2 and Smad3 at the C-terminal ends by the kinase activity of the type I receptor. The phosphorylated Smads then complex with Smad4, and the complex is then translocated into the nucleus. We used another gel shift experiment to determine whether the nucleus-localized Smad2-Smad4 and Smad3-Smad4 complexes activated by the TGF-␤ receptor attained the capacity to bind the SBE of Smad7 promoter. The Myc-tagged Smad2, Myc-tagged Smad3, and FLAG-tagged Smad4 were transfected into HEK293 cells in the presence or absence of a constitutively active TGF-␤ type I receptor, CA-ALK-5. The transfected cells were used to prepare nuclear extracts for a gel shift assay with the 33-bp SBE probe. As shown in Fig. 5A, expression of CA-ALK-5 did not induce a detectable SBE-binding complex in vector-transfected cells (lane 2 compared with lane 1), indicating that the endogenous level of the Smad2-Smad4 or Smad3-Smad4 complexes activated by the receptor might be too low to be detected under our experimental condition. Likewise, transfection of Smad2 or Smad4 alone could not lead to a detectable complex that binds the probe upon coexpression of CA-ALK-5 (lanes 3, 4, 7, and 8). However, expression of Smad3 in these cells was able to mediate the formation of a SBE-binding complex that was induced by co-expressed TGF-␤ type I receptor (lane 6 compared with lane 5), indicating that the activated receptor may aid in the nuclear translocation of Smad3 or the expressed Smad3 could cooperate with endogenous Smad4 to bind Smad7 SBE. When Smad2 was co-expressed with Smad4 in these cells, CA-ALK-5 clearly induced a SBE binding complex (lane 10 compared with lane 9) that contained Smad4, as detected by the supershift assay with an anti-FLAG antibody (lane 12). As expected, co-expression of Smad3 with Smad4 led to a receptor-inducible formation of a complex that binds Smad7 promoter (lane 14 compared with lane 13), and this complex contained both Smad3 (lane 15,

FIG. 3. Association of Smad3 and
Smad4 with Smad7 promoter. A, binding of Smad3 and Smad4 to a 282-bp Smad7 promoter fragment by gel mobility shift assay. The GST, GST-Smad2, GST-Smad3 (tagged with a Myc epitope), and GST-Smad4 (about 0.2 g/reaction) purified from bacteria were incubated with 32 P-labeled Smad7 promoter. Different cold competitor DNA (282 or 33 bp) or an anti-Myc antibody (about 1 g) were included in the binding reaction as indicated. The Smad3-mediated shift is marked by an arrow. B, association of Smad3 and Smad4 with the Smad binding element of the Smad7 promoter. Different GST fusion proteins were incubated with a 33-bp probe that contains the putative SBE of Smad7 promoter in a gel shift assay. Different amount of cold 33-bp competitor (10 -200 ng) or an anti-Myc antibody (0.05-1 g) were used as indicated.
FIG. 4. An intact Smad binding motif is required for TGF-␤ stimulation of the Smad7 promoter. HepG2 cells were transfected with Smad7 promoter/luciferase constructs Ϫ408 to Ϫ124 bp or Ϫ283 to Ϫ124 bp (0.25 g) and treated with or without activin A (10 ng/ml) or TGF-␤1 (1 ng/ml). The Ϫ124 to Ϫ283-bp construct changed the putative consensus Smad binding site GTCTAGAC of Smad7 promoter to CTCTAGAC. The fold change of luciferase activity was shown as the mean Ϯ S.D.
supershift with anti-Myc antibody) and Smad4 (lane 16, supershift with anti-FLAG antibody). These data clearly indicate that TGF-␤ signaling is able to induce the formation of Smad2-Smad4 and Smad3-Smad4 complexes in the nucleus and that these complexes are implicated in the regulation of Smad7 promoter through the direct interaction of Smad3 or Smad4 with the SBE of the Smad7 promoter. It is also interesting to note that the putative nuclear Smad2-Smad4 complex may behave differently from the Smad3-Smad4 complex in its interaction with the Smad7 promoter. We could not detect a convincing presence of Smad2 in the hypothetical SBE-binding Smad2-Smad4 complex by the gel shift assay (Fig. 5A, lane 11), although Smad2 expression was confirmed by anti-Myc Western blotting analysis with the same nuclear extracts (data not shown). This observation indicates that Smad4 may dissociate from Smad2 after the Smad2-Smad4 complex is translocated into the nucleus.
Our functional analysis with Smad7 promoter constructs has indicated that signaling with TGF-␤ and activin was able to activate the promoter activity in HepG2 cells (Fig. 2). To further confirm that this transactivation was directly related to the regulation of the SBE present in Smad7 promoter, we asked whether or not TGF-␤1 treatment in HepG2 cells was able to induce the formation of an endogenous nuclear protein complex that binds the SBE. HepG2 cells were treated with TGF-␤1 for 30, 60, and 120 min, and the nuclear extracts were used in a gel shift experiment with the 33-bp SBE probe. As shown in Fig. 5B, TGF-␤1 did induce the formation of a nuclear complex that associated with this SBE probe, and the associa-tion appeared to reach a maximum at 60 min after the treatment. Because this SBE was able to associate specifically with Smad3 and Smad4 (Figs. 3 and 5A), these data strongly suggest that TGF-␤ was able to induce the formation of an endogenous Smad complex that binds the SBE of Smad7 promoter.
In conclusion, our initial characterization of the Smad7 promoter provides one of the molecular mechanisms underlying the regulation of this inhibitory Smad by TGF-␤ and activin signaling at the transcriptional level. Treatment with TGF-␤ and activin in HepG2 cells was able to stimulate the activity of the Smad7 promoter. Furthermore, the Smad7 promoter contains a Smad binding element that is able to directly associate with Smad3 and Smad4. Activation of TGF-␤ signaling by the activated type I receptor was able to mediate the formation of a nuclear Smad3-Smad4 complex that bound the Smad7 promoter. These data clearly indicated that TGF-␤ and activin signaling may activate their cognate Smad proteins to transactivate the promoter of Smad7, which in turn blocks the signaling by these two extracellular factors. In addition to the function in a negative feedback loop that modulates TGF-␤ and activin signaling, regulation of the Smad7 message has been identified in other biological processes. The Smad7 mRNA level is up-regulated by laminar shear stress in vascular endothelial cells (45). Interferon-␥ is also able to stimulate the expression of Smad7 through JAK1 and STAT1 to block TGF-␤ signaling (46). With the cloned Smad7 promoter, it is now more feasible to address the question of how this inhibitory Smad is regulated by multiple signals at the transcriptional level.
FIG. 5. Activation of TGF-␤ signaling induced the formation of a nuclear complex that binds the SBE of Smad7 promoter. A, activated TGF-␤ type I receptor induced association of Smad3 and Smad4 with Smad7 promoter in HEK293 cells. Different combinations of Smad proteins or a constitutively active TGF-␤ type I receptor (CA-ALK-5) were transiently expressed in HEK293 cells. The nuclear extracts were prepared 48 h after the transfection and used in a gel shift assay with the 33-bp Smad7 promoter probe. The anti-Myc or anti-FLAG antibodies (about 1 g) were used in a supershift assay to detect the presence of Myc-tagged Smad2 and Smad3 and FLAG-tagged Smad4. B, TGF-␤1 treatment induced the formation of an endogenous SBE-binding nuclear complex in HepG2 cells. HepG2 cells were treated with TGF-␤1 (1 ng/ml) for 30, 60, and 120 min, and the nuclear extracts were used in a gel shift assay with the 33-bp SBE probe. The TGF-␤induced SBE-binding complex is indicated by an arrow.