Hoxa-9 Represses Transforming Growth Factor-β-induced Osteopontin Gene Transcription*

Smad2 and Smad3 are downstream transforming growth factor-β (TGF-β) signaling molecules. Upon phosphorylation by its type I receptor, Smad2 or Smad3 forms a complex with Smad4 and translocates to the nucleus where the complex activates target gene transcription. In the present study, we report that Smad3 binds directly to the osteopontin (OPN) promoter and that Smad4 interacts with the Hox protein and displaces it from its cognate DNA binding site in response to TGF-β stimulation. In gel shift assays, the glutathione S-transferase-Smad3 fusion protein was found to bind to a 50-base pair DNA element (−179 to −229) from the OPN promoter. Also, we found that both Hoxc-8 and Hoxa-9 bound to a Hox binding site adjacent to Smad3 binding sequence. Interestingly, Smad4, the common partner for both bone morphogenic protein and TGF-β signaling pathways, inhibited the binding of Hox protein to DNA. FLAG-tagged Smad4 coimmunoprecipitated with HA-tagged Hoxa-9 from cotransfected COS-1 cells, demonstrating an interaction between Smad4 and Hoxa-9. Transfection studies showed that Hoxa-9 is a strong transcriptional repressor; it suppresses the transcription of the luciferase reporter gene driven by a 124-base pair OPN promoter fragment containing both Smad3 and Hox binding sites. Taken together, these data demonstrate a unique TGF-β-induced transcription mechanism. Smad3 and Smad4 exhibit different functions in activation of OPN transcription. Smad3 binds directly to the OPN promoter as a sequence-specific activator, and Smad4 displaces the transcription repressor, Hoxa-9, by formation of Smad4/Hox complex as part of the transcription mechanism in response to TGF-β stimulation.

stimulates the proliferation of precursor cells of osteoblast lineage, which induces new bone formation (1,2). TGF-␤ decreases bone resorbtion by inhibiting both proliferation and differentiation of osteoclast precursors (3,4) and by inducing apoptosis of osteoclasts (5). Osteopontin (OPN), the major noncollagenous bone matrix protein, is a secreted, arginine-glycine-aspartate (RGD)-containing phosphorylated glycoprotein (6,7). OPN expression is rapidly induced by both bone morphogenic protein (BMP) and TGF-␤ (8,9). It is produced by osteoblasts (10), as well as osteoclasts (10,11). OPN regulates the adhesion, attachment, and spreading of osteoclasts to the bone surface during bone resorbtion (12). Most recently, OPN has been demonstrated to regulate the bone remodeling in response to mechanical stress (13), a novel mechanism by which osteoclast function is related to mechanical loading of the bone tissue.
Signal transduction by members of TGF-␤ superfamily is mediated by two types of transmembrane receptors (14). Smads are the direct substrates for kinase receptors. Upon phosphorylation, Smads translocate to the nucleus and recruit DNAbinding protein(s) to regulate gene expression. Smads are functionally classified into three groups: receptor-regulated Smads (R-Smads) including Smad1, -5, and -8 for BMP (15)(16)(17)(18)(19) and Smad2 and -3 for TGF-␤/activin signaling pathway (20 -23); co-Smad Smad4, which hetero-oligomerizes with R-Smads (24 -29); and anti-Smads, Smad6 and -7, which block signals from being transduced into the nucleus via a mechanism of associating with type I kinase receptors (30 -33) or competing with Smad4 for pathway-specific Smads (34). The activated type I receptors phosphorylate specific R-Smads. Upon dissociation of the phosphorylated R-Smad from type I receptor, it interacts with Smad4 and moves to the nucleus as a heteromeric complex. Once in the nucleus, Smads can be recruited to many DNA-binding proteins such as c-Jun/c-Fos, Fast-1, Fast-2, vitamin D receptor, glucocorticoid receptor, and Hoxc-8 to regulate transcriptional responses (24,35,37). It is not clear, however, whether the Smad4-R-Smads complex still remains associated after translocating to the nucleus (38,39).
In a previous study (40), we have demonstrated an interaction between Smad1 and Hoxc-8 in the BMP signaling pathway. The interaction between Smad1 and Hoxc-8 breaks the equilibrium of Hoxc-8 binding to its DNA binding site and results in the transcriptional activation of OPN in response to BMP2 stimulation (40). Here we show evidence that the DNAbinding protein Hoxa-9 interacts with Smad4, but not with Smad3 (which binds to OPN promoter), and the interaction between Smad4 and Hoxa-9 results in the transcriptional activation of OPN in response to TGF-␤ stimulation. Also, unlike most DNA-binding proteins interacting with Smads that are transcriptional activators, Hoxa-9 functions as a strong transcriptional repressor, similar to Hoxc-8. * This work was supported in part by an Arthritis Foundation Investigator Grant (to X. S.), National Institutes of Health Grant DK53757 (to X. C.), and Department of Army Grant DAMD17-00-1-0066 (to X. C.). 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.

EXPERIMENTAL PROCEDURES
Cell Lines-Mink lung epithelial cell line Mv1Lu (ATCC, Manassas, VA) was maintained in minimal essential medium containing 10% fetal bovine serum and nonessential amino acids. COS-1 (ATCC) cells were maintained in Dulbecco's modified Eagle's medium supplied with 10% fetal bovine serum.
Plasmid Constructs-OPN promoter luciferase reporter constructs Hox-pGL3 and mHox-pGL3, bacteria expression vectors for GST-Smad2, GST-Smad-3, GST-Smad-4, GST-Hoxa-9, as well as mammalian expression vectors Smad3-FLAG and Smad4-FLAG were described previously (40). mSBE-pGL3, which bears four nucleotide alterations at the Smad binding site in the OPN promoter Ϫ168 to Ϫ229 region, was generated by standard polymerase chain reaction cloning technique using primers 5Ј-TATAACGCGTCTAAATGCCATGGATAAATGAAA-AGG-3Ј (upstream) and 5Ј-TATACTCGAGTACACAAAGCATTACTG-A-3Ј(downstream) and inserted in between MluI and XhoI sites of the pGL3-control vector (Promega). An MluI and an XhoI site were added to the up-and downstream primers (bold), respectively. Mutated nucleotides are shown underlined. HA-Hoxa-9 expression vector was constructed in a similar strategy using a full-length Hoxa-9 cDNA vector (kindly provided by C. Largman) as a template. The primer sequences for up-and downstream, respectively, are: 5Ј-TATAGGATCCATGGC-CACCACCGGGGCCCTGGGCAA-3Ј and 5Ј-TATATCTAGACGGACA-GTCCTTTCTTTTTCTTGTCT-3Ј. A BamHI and an XbaI restriction site (bold) was added to the upstream and downstream primers, respectively. The polymerase chain reaction product was cut with XbaI and inserted between the EcoRI (blunted with S1 nuclease) and XbaI sites of the PcDNA3-HA-Hoxc-8 (40).
Expression and Purification of GST Fusion Proteins-BL21 bacterial cells were transformed with the GST fusion constructs described above, and the fusion proteins were purified by glutathione-agarose 4B (Sigma) following induction with isopropyl-␤-D-thiogalactopyranoside as described previously (41).
Electrophoretic Mobility Shift Assays-A 50-bp SBE (Smad binding element) corresponding to nucleotides Ϫ 179 to Ϫ229, and a 25-bp Smad binding element (used in Fig. 4) corresponding to nucleotides Ϫ203 to Ϫ228 of the OPN promoter region were generated by annealing pairs of oligonucleotides (sense strands are shown). 5Ј-CTAAATGCAGTC-TATAAATGAAAAGGGTAGTTAATGACATCGTTCATCAG-3Ј for SBE (underlined sequences were changed to ACCCTT and GCGC, respectively, in mutant probe m-SBE), and 5Ј-CTAAATGCAGTCTATAAAT-GAAAAG-3Ј for Smad binding element. The probe OPN5 used in Fig. 5 was described previously (40). These DNA fragments were end-labeled with [␥-32 P]ATP using T4 polynucleotide kinase. The subsequent experiment was performed as described previously (41). For supershift assays, anti-FLAG M2 antibody (Eastman Kodak Co.) was added to the binding reaction and incubated for an additional 10 min at room temperature.
Immunoprecipitation and Western Blotting-COS-1 cells were transfected with expression vectors as indicated in Fig. 6 using Tfx-50 according to manufacturer's description (Promega). Twenty-four hours post-transfection, the cells were switched to Dulbecco's modified Eagle's medium containing 0.2% fetal bovine serum for 12 h and then treated with 10 ng/ml TGF-␤1 (R & D Systems, Minneapolis, MN) for 30 min where appropriate. The cell lysates were immunoprecipitated with anti-HA polyclonal antiserum (Babco, Berkeley, CA) and immunoblotted with anti-FLAG M2 monoclonal antibody (Kodak) as described previously (41). For the gel shift assay (see Fig. 4), the cells were treated with 10 ng/ml TGF-␤1 for 12 h before cells were lysed.
Transfection and Luciferase Assays-Mv1Lu cells were transiently transfected with the constructs indicated in Fig. 1 using a mixture of cationic and neutral lipids (Tfx-50, Promega) as described previously (41). When increasing amounts of expression vectors were transfected, total DNA was kept constant by addition of PcDNA3 (Invitrogen, Carlsbad, CA). An internal control plasmid pRL-SV40 was cotransfected to monitor the transfection efficiency. Luciferase activities were assayed 48 h post-transfection with separate substrates to detect the luciferase (firefly) in the promoter-luciferase reporter plasmid and to the second luciferase (Renilla), encoded by the pRL-SV40 vector (dual luciferase assay kit, Promega) according to the manufacturer's directions. Values were normalized to the renilla luciferase activity expressed from the pRL-SV40 reporter plasmid. Luciferase values shown in the figures are representative of transfection experiments performed in triplicate in at least three independent experiments.

RESULTS AND DISCUSSION
TGF-␤ Activates OPN Gene Transcription-To investigate the molecular basis of TGF-␤-induced OPN expression, the effect of TGF-␤ on OPN promoter activity was tested. Hox-pGL3 (Fig. 1A), a previously characterized OPN promoterluciferase reporter (40), was used. Hox-pGL3 contains a 124-bp OPN promoter fragment (nucleotides Ϫ166 to Ϫ290) inserted upstream of SV40 promoter in the pGL3-control vector. This 124-bp fragment contains a SBE (nucleotides Ϫ217 to Ϫ222) and a well characterized Hox binding site (TAAT, nucleotides Ϫ195 to Ϫ198) separated by 20 base pairs. Mv1Lu cells, a TGF-␤-responsive cell line, were transiently transfected with Hox-pGL3 reporter construct. The transfected cells were then incubated in the presence or absence of 5 ng/ml TGF-␤ for 48 h before the luciferase activity was assayed. As shown in Fig. 1B, treatment of transfected cells with TGF-␤ stimulated OPN promoter-reporter activity more than 5-fold. To verify whether SBE is involved in the TGF-␤-induced transcriptional activity, a shorter, mutant version of OPN promoter-luciferase reporter was constructed (nucleotides Ϫ168 to Ϫ229) in pGL3-control vector (Fig. 1A). This 61-bp OPN promoter reporter construct (mSBE-pGL3) bears four nucleotide alterations in the Smad binding sequence (CAGTCT to CCATGG). Transfection study showed that mutation of the Smad binding site greatly reduced the basal and TGF-␤-induced reporter activity. These results suggested that the TGF-␤-induced OPN gene transcription is mediated via the SBE and that the direct binding of Smad3 to OPN promoter might be required for its gene activation.
binding element was examined in gel shift assays. As shown in Fig. 2, GST-Hoxa-9 bound to this DNA element (lane 3). The specificity of the binding was demonstrated by competition shift assays using unlabeled specific probe (lanes 4 -6), as well as nonspecific probe (lanes 7-9). The GST itself did not bind to this probe (lane 2). Lane 1 is the probe with no added protein.
These results suggest that the inhibitory effect of Hoxa-9 on the OPN promoter is mediated via the Hox binding site.
Smad3 Binds to the OPN Promoter-We then investigated if the direct binding of Smad2 or Smad3 to OPN promoter is required. Gel shift assays were performed using affinity-purified GST-Smad2 or GST-Smad3 fusion protein and a 50-bp DNA fragment containing SBE (Ϫ179 to Ϫ229). Fig. 3A shows that GST-Smad3 effectively bound to SBE (lane 3), while GST-Smad2 bound weakly (data not shown). Thus, we focused attention on Smad3. The specificity of Smad3 binding was shown in a competition shift assay. Unlabeled SBE dose-dependently competed for Smad3 binding (lanes 4 -6), and a 100-fold excess of it completely competed off Smad3 binding (lane 6). In contrast, an equal molar amount of a nonspecific probe did not show such an effect (lanes 7-9). GST-Smad4, the common partner for all pathway-specific Smads, did not bind to SBE (lane 10), nor did it affect Smad3 binding activity (lanes 11-13). To validate the DNA sequence in the SBE that confers the binding of Smad3, the core sequences CAGTCT for SBE and TAAT for Hox protein binding sites were mutated to ACCCTT and GCCG, respectively. As shown in Fig. 3B, mutations of SBE abolished binding of Smad3 (Fig. 3B, compare lane 7 to wildtype probe, lane 3).
Studies have shown that the MH1 domain, but not the full length of Smad3, is able to bind DNA (41,42). Under our experimental conditions, however, full-length GST-Smad3 was found to bind to OPN promoter effectively (Fig. 3). To confirm that the full length of mammalian cell-expressed Smad3 also bound to the SBE, COS-1 cells were transfected with Smad3-FLAG expression vector. The transfected cells were treated with or without TGF-␤ (10 ng/ml) for 12 h before lysates were prepared, then gel shift assays were performed using the SBE probe. Western blot demonstrated that Smad3 was only expressed in the cells transfected with the Smad expression plasmid (Fig. 4B). As shown in Fig. 4A, a shifted band is seen from both TGF-␤-treated and untreated lysates of Smad3-FLAG transfected cells (lanes 4 and 5). In contrast, the lysates from control cells did not yield such a band (lanes 2 and 3). To verify the presence of FLAG-tagged Smad3 in this new complex, anti-FLAG M2 antibody was added to the binding reaction. The addition of antibodies disrupted the newly formed DNA-protein complex (lanes 8 and 9), indicating that the complex contained FLAG-tagged Smad3. Lane 1 contained no protein, demonstrating the quality of the probe.
In summary, we have located a 50-bp DNA fragment (Ϫ179 to Ϫ229) in the OPN promoter to be responsible for mediating TGF-␤-induced OPN gene transcription. It is interesting to note that this 50-bp fragment contains a functional Smad3 and a Hox binding sites. Mutations of these two sites abolished both Smad3 and Hoxa-9 binding activities (Fig. 3B). Disruption of the Smad binding site reduced basal promoter activity significantly, but did not abolish the TGF-␤-induced OPN gene transcription (Fig. 1). This is likely due to the interactions of endogenous Smad4 with Hox proteins, since the endogenous Smad2 and -3 will associate with Smad4 and take it into the nucleus upon TGF-␤ stimulation where Smad4, in turn, interacts with and dislodges Hox proteins from their cognate bind-ing sites. As shown in our previous studies, mutation of the Hox binding site in Hox-pGL3 reporter construct elevates the basal promoter activity and de-represses the inhibitory effects of Hox proteins on this promoter-luciferase reporter (40).
Our previous studies have shown that Hoxc-8 is the downstream transcription factor of the BMP signaling pathway and that BMP-2 activates OPN gene transcription through an interaction between Smad1 and Hoxc-8. The results of this study show that Hox binding element is also involved in the TGF-␤induced OPN gene transcription via an interaction between Hoxa-9 and Smad4 and that Smad3 directly binds to the OPN promoter in response to TGF-␤. It is well established that phosphorylated Smad3 associates with Smad4 and translocates to nucleus as a complex. Smad4 is not required for nuclear accumulation but for the formation of functional complexes. However, it is not clear whether Smad4 and Smad3 still remain as complex in participating transcription after translocation to nucleus. Our data indicate that Smad4 and Smad3 can function separately in activation of OPN gene transcription. While Smad4 displaces transcriptional repressor, Hoxa-9, from its binding site, Smad3 directly binds to OPN promoter as a sequence specific activator.