Voltar
Artigo Acesso aberto Revisado por pares
BIBTEX RIS

Interaction between GC Box Binding Factors and Smad Proteins Modulates Cell Lineage-specific α2(I) Collagen Gene Transcription

2001; Elsevier BV; Volume: 276; Issue: 19 Linguagem: Inglês

10.1074/jbc.m010485200

ISSN

1083-351X

Autores

Yutaka Inagaki, Tomoyuki Nemoto, Atsuhito Nakao, Peter ten Dijke, Kenichi Kobayashi, Kazuhiko Takehara, Patricia Greenwel,

Tópico(s)

Liver physiology and pathology

Resumo

Type I collagen is produced predominantly in mesenchymal cells, but molecular mechanisms responsible for cell type-specific expression are virtually unknown. During fibrogenic process in the liver, activated hepatic stellate cells (HSC) are the main producers of type I collagen, whereas parenchymal hepatocytes produce little, if any, of this protein. We have previously reported that Sp1 and an interacting unknown factor(s) bind to the −313 to −255 sequence of the α2(I) collagen gene (COL1A2) and play essential roles for basal and TGF-β-stimulated transcription in skin fibroblasts and HSC. Recently, Smad3 has been shown to bind to this region, and its interaction with Sp1 has been implicated in TGF-β-elicited COL1A2 stimulation. The present study demonstrates predominant binding of Sp3 rather than Sp1 to this regulatory element in parenchymal hepatocytes. In these cells, this region did not exhibit strong enhancer activity or mediate the effect of TGF-β. Transfection of HSC with an Sp3 expression plasmid abolished the COL1A2 response to TGF-β, whereas overexpression of Sp1 in hepatocytes increased basal COL1A2transcription and conferred TGF-β responsiveness. Functional and physical interactions between Sp1 and Smad3, but not between Sp3 and Smad3, were demonstrated using the bacterial GAL4 system and immunoprecipitation-Western blot analyses. These results indicate that cell lineage-specific interactions between GC box binding factors and Smad protein(s) may account, at least in part, for differential COL1A2 transcription and TGF-β responsiveness in HSC and parenchymal hepatocytes. Type I collagen is produced predominantly in mesenchymal cells, but molecular mechanisms responsible for cell type-specific expression are virtually unknown. During fibrogenic process in the liver, activated hepatic stellate cells (HSC) are the main producers of type I collagen, whereas parenchymal hepatocytes produce little, if any, of this protein. We have previously reported that Sp1 and an interacting unknown factor(s) bind to the −313 to −255 sequence of the α2(I) collagen gene (COL1A2) and play essential roles for basal and TGF-β-stimulated transcription in skin fibroblasts and HSC. Recently, Smad3 has been shown to bind to this region, and its interaction with Sp1 has been implicated in TGF-β-elicited COL1A2 stimulation. The present study demonstrates predominant binding of Sp3 rather than Sp1 to this regulatory element in parenchymal hepatocytes. In these cells, this region did not exhibit strong enhancer activity or mediate the effect of TGF-β. Transfection of HSC with an Sp3 expression plasmid abolished the COL1A2 response to TGF-β, whereas overexpression of Sp1 in hepatocytes increased basal COL1A2transcription and conferred TGF-β responsiveness. Functional and physical interactions between Sp1 and Smad3, but not between Sp3 and Smad3, were demonstrated using the bacterial GAL4 system and immunoprecipitation-Western blot analyses. These results indicate that cell lineage-specific interactions between GC box binding factors and Smad protein(s) may account, at least in part, for differential COL1A2 transcription and TGF-β responsiveness in HSC and parenchymal hepatocytes. genes coding for the α1 and α2 chains of type I collagen, respectively chloramphenicol acetyltransferase CCAAT/enhancer-binding protein(s) cytomegalovirus fetal bovine serum hepatic stellate cells TGF-β-responsive element transforming growth factor-β hemagglutinin Type I collagen, the major component of extracellular matrix in various organs, is produced predominantly in mesenchymal cells such as fibroblasts, osteoblasts, and myofibroblasts. It is composed of two α1 chains and one α2 chain, which are coordinately expressed but encoded by the distinct genes,COL1A1 1 andCOL1A2, respectively. Transforming growth factor-β1 (TGF-β1, henceforth referred to as TGF-β) plays important roles in stimulating type I collagen gene expression mainly at the levels of transcription (1Ramirez F. Di Liberto M. FASEB J. 1990; 4: 1616-1623Crossref PubMed Scopus (88) Google Scholar). In the liver, it has been widely accepted that activated hepatic stellate cells (HSC) showing the morphological and functional features of myofibroblasts are the main producers of type I collagen (2Friedman S.L. Seminars Liver Dis. 1990; 10: 20-29Crossref PubMed Scopus (188) Google Scholar, 3Maher J.J. McGuire R.F. J. Clin. Invest. 1990; 86: 1641-1648Crossref PubMed Scopus (364) Google Scholar). By contrast, parenchymal hepatocytes produce little, if any, collagen during hepatic fibrogenesis (4Takahara T. Kojima T. Miyabayashi C. Inoue K. Sasaki H. Muragaki Y. Ooshima A. Lab. Invest. 1988; 59: 509-521PubMed Google Scholar, 5Milani S. Herbst H. Schuppan D. Hahn E.G. Stein H. Hepatology. 1989; 10: 84-92Crossref PubMed Scopus (261) Google Scholar). Recent studies by us and others have indicated that both COL1A1 (6Houglum K. Buck M. Alcorn J. Contreras S. Bornstein P. Chojkier M. J. Clin. Invest. 1995; 96: 2269-2276Crossref PubMed Scopus (47) Google Scholar) and COL1A2 (7Inagaki Y. Truter S. Bou-Gharios G. Garrett L.A. de Crombrugghe B. Nemoto T. Greenwel P. Biochem. Biophys. Res. Commun. 1998; 250: 606-611Crossref PubMed Scopus (33) Google Scholar) transgenes are expressed in mesenchymal cells, but not in parenchymal hepatocytes, after carbon tetrachloride administration into transgenic mice harboring collagen promoter-reporter gene constructs. These results further confirmed the minor contribution of hepatocytes to collagen production. However, very little is known regarding the molecular mechanisms determining differential type I collagen gene expression in activated HSC and parenchymal hepatocytes. We have previously shown that similar regulatory mechanisms controlCOL1A2 transcription in skin fibroblasts (8Boast S. Su M.-W. Ramirez F. Sanchez M. Avvedimentro E.V. J. Biol. Chem. 1990; 265: 13351-13356Abstract Full Text PDF PubMed Google Scholar, 9Inagaki Y. Truter S. Ramirez F. J. Biol. Chem. 1994; 269: 14828-14834Abstract Full Text PDF PubMed Google Scholar) and activated HSC (10Inagaki Y. Truter S. Greenwel P. Rojkind M. Unoura M. Kobayashi K. Ramirez F. Hepatology. 1995; 22: 573-579PubMed Google Scholar). The COL1A2 upstream sequence spanning from −313 to −183 is essential for basal transcription of the gene in the two cell types of mesenchymal origin and mediates stimulatory effect of TGF-β on COL1A2 transcription. Thus, we designated this region the TGF-β-responsive element (TbRE). Within the TbRE there are at least two distinct sites of DNA-protein interaction, Box 3A (−313 to −286) and Box B (−271 to −255; Fig.1). Box 3A contains two GC boxes, which are bound by a ubiquitous trans-activator Sp1 in skin fibroblasts and HSC (9Inagaki Y. Truter S. Ramirez F. J. Biol. Chem. 1994; 269: 14828-14834Abstract Full Text PDF PubMed Google Scholar). An additional GC box is present in the intervening sequence between Box 3A and Box B. Our previous work has demonstrated that TGF-β up-regulates COL1A2 transcription by modifying the interaction between Box 3A-bound Sp1 and an unknown nuclear factor(s) bound to the neighboring Box B sequence (9Inagaki Y. Truter S. Ramirez F. J. Biol. Chem. 1994; 269: 14828-14834Abstract Full Text PDF PubMed Google Scholar, 11Greenwel P. Inagaki Y. Hu W. Walsh M. Ramirez F. J. Biol. Chem. 1997; 272: 19738-19745Abstract Full Text Full Text PDF PubMed Scopus (173) Google Scholar). Recently, it has been shown that Smad3, an intracellular mediator of TGF-β signal transduction, binds to the CAGACA sequence present in Box B and stimulates COL1A2 transcription (12Chen S.-J. Yuan W. Mori Y. Levenson A. Trojanowska M. Varga J. J. Invest. Dermatol. 1999; 112: 49-57Abstract Full Text Full Text PDF PubMed Scopus (377) Google Scholar, 13Chen S.-J. Yuan W. Lo S. Trojanowska M. Varga J. J. Cell. Physiol. 2000; 183: 381-392Crossref PubMed Scopus (171) Google Scholar). Furthermore, we have recently demonstrated that a functional interaction between Sp1 and Smad3/Smad4 is critical for TGF-β-elicited stimulation of COL1A2 transcription in fibroblasts (14Zhang W. Oul J. Inagaki Y. Greenwel P. Ramirez F. J. Biol. Chem. 2000; 275: 39237-39245Abstract Full Text Full Text PDF PubMed Scopus (214) Google Scholar), whereas others have implicated functional cooperation with p300/CBP in Smad-dependent stimulation ofCOL1A2 transcription (15Ghosh A.K. Yuan W. Mori Y. Varga J. Oncogene. 2000; 19: 3546-3555Crossref PubMed Scopus (205) Google Scholar). The present study was designed to determine whether collagen gene transcription is differentially regulated in activated HSC and parenchymal hepatocytes and, if so, to study the molecular mechanisms responsible for this cell type-specific expression. Our results indicated that, unlike in mesenchymal cells such as skin fibroblasts and HSC, strong enhancer activity of the −313 to −183 segment and response to TGF-β were not observed in primary culture of hepatocytes. Gel mobility shift assays revealed that, whereas Sp1 was the major Box 3A-bound factor in HSC, Sp3 bound predominantly to this GC-rich sequence in hepatocytes. Transfection of activated HSC with an Sp3 expression plasmid abolished the COL1A2 response to TGF-β. By contrast, overexpression of Sp1 in hepatocytes increased the basal level of COL1A2 transcription and conferred TGF-β responsiveness. Functional and physical interactions between Sp1 and Smad3, but not between Sp3 and Smad3, were demonstrated using the bacterial GAL4 fusion protein system and immunoprecipitation-Western blot analyses. Based on these lines of experimental evidence, we conclude that interaction between Sp1/Sp3 transcription factors and Smad proteins modulates, at least in part, cell lineage-specific COL1A2 transcription in HSC and parenchymal hepatocytes. Adult rat hepatocytes and HSC were prepared by collagenase perfusion (16Summerfield J.A. Vergalla J. Jones E.A. J. Clin. Invest. 1982; 69: 1337-1347Crossref PubMed Scopus (58) Google Scholar) and Pronase-collagenase perfusion (17Greenwel P. Schwartz M. Rosas M. Peyrol S. Grimaud J.-A. Rojkind M. Lab. Invest. 1991; 65: 644-653PubMed Google Scholar), respectively. They were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (for hepatocytes) or 20% (for HSC) fetal bovine serum (FBS). More than 95% purity of parenchymal hepatocytes and more than 90% purity of HSC were confirmed immunohistochemically using anti-albumin and anti-desmin antibodies, respectively (data not shown). Primary HSC passaged one to three times were used for transfection experiments. CFSC-2G is an activated HSC clone derived from a cirrhotic rat liver induced by carbon tetrachloride injection (18Greenwel P. Rubin J. Schwartz M. Hertzberg E.L. Rojkind M. Lab. Invest. 1993; 69: 210-217PubMed Google Scholar), and the cells were maintained in the same medium supplemented with 10% FBS and non-essential amino acids. Because primary culture of rat hepatocytes were found to maintain their differentiated phenotype better on type I collagen-coated dishes (Corning Glass Works, Corning, NY; data not shown), primary HSC and CFSC-2G cells were plated on the same substrata for a strict comparison in the experiments shown in Fig.2. COS-7 cells were obtained from the American Type Culture Collection and were cultured in Dulbecco's modified Eagle's medium with 10% FBS. Plasmids containing different lengths of COL1A2 upstream sequence linked to either a bacterial chloramphenicol acetyltransferase (CAT) gene or a firefly luciferase gene have been previously described (8Boast S. Su M.-W. Ramirez F. Sanchez M. Avvedimentro E.V. J. Biol. Chem. 1990; 265: 13351-13356Abstract Full Text PDF PubMed Google Scholar, 9Inagaki Y. Truter S. Ramirez F. J. Biol. Chem. 1994; 269: 14828-14834Abstract Full Text PDF PubMed Google Scholar, 19Inagaki Y. Truter S. Tanaka S. Di Liberto M. Ramirez F. J. Biol. Chem. 1995; 270: 3353-3358Abstract Full Text Full Text PDF PubMed Scopus (105) Google Scholar) and are schematically shown in Fig. 1. A promoterless CAT construct, pBLCAT3 (20Luckow B. Schütz G. Nucleic Acids Res. 1987; 15: 5490Crossref PubMed Scopus (1401) Google Scholar), was used as a negative control. An SV40 early promoter region was cloned into pBLCAT3 vector (designated pSVCAT3) and used as a positive control. Expression plasmids used are pCMV-Sp1 (21Elder G.A. Liang Z. Li C. Lazzarini R.A. Nucleic Acids Res. 1992; 20: 6281-6285Crossref PubMed Scopus (10) Google Scholar) kindly provided by Dr. G. Elder, pCMV-Sp3 (22Murata Y. Kim H.G. Rogers K.T. Udvadia A.J. Horowitz J.M. J. Biol. Chem. 1994; 269: 20674-20681Abstract Full Text PDF PubMed Google Scholar) from Dr. J. Horowitz, and pCMV-Smad3 (23Nakao A. Imamura T. Souchelnytskyi S. Kawabata M. Ishisaki A. Oeda E. Tamaki K. Hanai J. Heldin C.-H. Miyazono K. ten Dijke P. EMBO J. 1997; 16: 5353-5362Crossref PubMed Scopus (919) Google Scholar) from Dr. R. Derynck. In these plasmids, either Sp1, Sp3, or Smad3 is expressed under the control of cytomegalovirus (CMV) promoter. An empty CMV-driven expression vector, pcDNA3 (Invitrogen Corp., Carlsbad, CA), was used as a negative control. Bacterial GAL4 fusion protein expression plasmids, pMSp1, pMDNSp1, and pMSp3, which encode the active form Sp1, transactivation domain-deleted Sp1, and active form Sp3, respectively, were generously provided by Dr. Y. Sowa, as well as a control pM plasmid and the pG5-luc reporter construct (24Sowa Y. Orita T. Minamikawa-Hiranabe S. Mizuno T. Nomura H. Sakai T. Cancer Res. 1999; 59: 4266-4270PubMed Google Scholar). Preparation of plasmid DNA for cell transfection was previously described (8Boast S. Su M.-W. Ramirez F. Sanchez M. Avvedimentro E.V. J. Biol. Chem. 1990; 265: 13351-13356Abstract Full Text PDF PubMed Google Scholar). CFSC-2G cells were transfected using the calcium phosphate coprecipitation technique (8Boast S. Su M.-W. Ramirez F. Sanchez M. Avvedimentro E.V. J. Biol. Chem. 1990; 265: 13351-13356Abstract Full Text PDF PubMed Google Scholar) followed by a 15% glycerol shock for 90 s. FuGENE 6 transfection reagent (Roche Diagnostic Co., Indianapolis, IN) was used for transfection of primary culture of rat hepatocytes and HSC. In some experiments, transfected cells were placed in medium containing 0.1% FBS and treated with 2 ng/ml of TGF-β (Collaborative Biomedical Products, Bedford, MA). Cells were harvested 48 h after transfection and subjected to either CAT or luciferase assays. Enzyme activity of the COL1A2·CAT chimeric constructs was normalized against that of co-transfected pSVXP1, in which the SV40 early promoter region was cloned into a promoterless luciferase gene construct, pXP1 (25Nordeen S.K. BioTechniques. 1988; 6: 454-457PubMed Google Scholar). Transcriptional activity of the COL1A2/luciferase chimeric constructs and pG5-luc reporter construct was normalized against that of co-transfected pRLCMV (Promega, Madison, WI), in which a Renilla luciferase gene was driven by a CMV promoter. To avoid competition between the CMV promoter region of the expression plasmids and that of pRLCMV vector, the latter DNA was added to the plasmid mixture at 1:1,000 molar ratio against the former. CAT assays and the dual luciferase assays were carried as previously reported (26Gorman C.M. Moffat L.F. Howard B.H. Mol. Cell. Biol. 1982; 2: 1044-1051Crossref PubMed Scopus (5293) Google Scholar) and according to the manufacturer's protocol (Promega), respectively. Nuclear extracts were prepared from culture cells as previously reported (27Truter S. Di Liberto M. Inagaki Y. Ramirez F. J. Biol. Chem. 1992; 267: 25389-25395Abstract Full Text PDF PubMed Google Scholar). After incubating nuclear extracts with an end-labeled probe, gel mobility shift assays were carried out as previously described (27Truter S. Di Liberto M. Inagaki Y. Ramirez F. J. Biol. Chem. 1992; 267: 25389-25395Abstract Full Text PDF PubMed Google Scholar). Binding conditions, as well as the sequences of the Box 5A (−330 to −297) and Box 3A (−313 to −286) oligonucleotides used as either probes or unlabeled competitors, have been previously described (9Inagaki Y. Truter S. Ramirez F. J. Biol. Chem. 1994; 269: 14828-14834Abstract Full Text PDF PubMed Google Scholar). Identical amounts of nuclear proteins from each cell source were added to the binding reactions. For antibody interference assays, antibodies against CCAAT/enhancer-binding proteins (C/EBPα, C/EBPβ, and C/EBPγ), NF1, or Sp1-related factors (Sp1, Sp2, Sp3, and Sp4) were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Immunoblotting of nuclear proteins was performed as previously described (11Greenwel P. Inagaki Y. Hu W. Walsh M. Ramirez F. J. Biol. Chem. 1997; 272: 19738-19745Abstract Full Text Full Text PDF PubMed Scopus (173) Google Scholar) using antibodies against Sp1 and Sp3. To analyze interactions between Smad3 and Sp1/Sp3, COS-7 cells were transfected with a Myc-tagged Smad3 expression plasmid (23Nakao A. Imamura T. Souchelnytskyi S. Kawabata M. Ishisaki A. Oeda E. Tamaki K. Hanai J. Heldin C.-H. Miyazono K. ten Dijke P. EMBO J. 1997; 16: 5353-5362Crossref PubMed Scopus (919) Google Scholar) together with an expression vector encoding either Sp1 or Sp3, in the presence or absence of HA-tagged constitutive active TGF-β type I receptor (ALK5TD) (23Nakao A. Imamura T. Souchelnytskyi S. Kawabata M. Ishisaki A. Oeda E. Tamaki K. Hanai J. Heldin C.-H. Miyazono K. ten Dijke P. EMBO J. 1997; 16: 5353-5362Crossref PubMed Scopus (919) Google Scholar). Forty-eight hours later, whole cell lysates were subjected to immunoprecipitation with anti-Myc antibodies (Santa Cruz Biotechnology, Inc.), followed by immunoblotting with either anti-Sp1 or anti-Sp3 antibodies. Some cell lysates from the transfected cells were directly immunoblotted with anti-Myc, anti-Sp1, anti-Sp3, or anti-HA antibodies (Roche Diagnostic Co.) to confirm expression of Myc-tagged Smad3, Sp1, Sp3, and HA-tagged ALK5TD, respectively, in the cells. Values were expressed as mean ± S.D. Either Student's t test or the Mann-WhitneyU test was used to evaluate the statistical differences between groups: a p value of less than 0.05 was considered significant. We first compared COL1A2 transcription and response to TGF-β between activated HSC (CFSC-2G) and primary culture of rat hepatocytes. Normalization of the assay data against the activity of the co-transfected pSVXP1 vector allowed us to make a comparison between the two cell types, despite the fact that we used two different methods of transfection. In primary culture of hepatocytes, transcriptional activity of the −378 and −313COL1A2·CAT constructs containing the TbRE was one-third of that in CFSC-2G transfectants (Fig. 2 A). Deletion of the −378 to −183 sequence did not affect COL1A2 transcription in hepatocytes: there were no significant differences in the CAT enzyme activity among the −378, −313, and −183COL1A2·CAT transfectants (Fig. 2 A). This was apparently different from the results obtained with CFSC-2G cells showing that transcriptional activity of the −378 and −313COL1A2·CAT constructs was significantly higher than that of the −183COL1A2·CAT construct (Fig. 2 A). Administration of 2 ng/ml of TGF-β into the culture medium resulted in a significant increase in transcriptional activity of the −378 and −313COL1A2·CAT constructs in CFSC-2G cells, but not in primary culture of hepatocytes (Fig. 2 A). Consistent with these results, a parallel experiment of Northern blot hybridization did not show any detectable amount of COL1A2 mRNA in either untreated or TGF-β-treated hepatocytes (data not shown). Transcription of the −183COL1A2·CAT construct did not show TGF-β responsiveness when transfected into either CFSC-2G cells or primary culture of hepatocytes (Fig. 2 A). Similar results were obtained when comparing COL1A2transcription between early passaged HSC, instead of an immortalized HSC clone CFSC-2G, and primary culture of hepatocytes (Fig.2 B). Because the above functional assays revealed a major difference in the promoter activity and TGF-β responsiveness of the −378 to −183 segment between HSC and primary culture of hepatocytes, we next examined the binding of transcription factors present in the nuclei of these cells to theCOL1A2 upstream sequence. Nuclear extracts prepared from primary culture of hepatocytes and CFSC-2G cells exhibited similar gel shift patterns when using the Box 5A oligonucleotide as a probe, although the intensities and relative ratios of the retarded bands were somewhat different from each other (Fig.3 A). By contrast, gel mobility shift assays using the Box 3A probe indicated that nuclear proteins prepared from primary culture of hepatocytes and CFSC-2G cells demonstrated slightly different migrating patterns. The faster migrating complexes obtained with both cell types showed different mobilities from each other (Fig. 3 A). Antibody interference assays revealed that anti-C/EBPβ antibodies effectively diminished the intensity of the Box 5A-bound complexes, particularly the fast migrating bands, in both primary culture of hepatocytes and CFSC-2G cells (Fig. 3 B), as previously reported with NIH3T3 fibroblasts (28Greenwel P. Tanaka S. Penkov D. Zhang W. Olive M. Moll J. Vinson C. Di Liberto M. Ramirez F. Mol. Cell. Biol. 2000; 20: 912-918Crossref PubMed Scopus (145) Google Scholar). None of the anti-C/EBPα, anti-C/EBPγ, and anti-NF1 antibodies interfered with the complex obtained with either hepatocytes or CFSC-2G cells (Fig. 3 B). When using the Box 3A oligonucleotide as a probe, anti-Sp1 antibodies completely diminished the slowly migrating complex obtained with CFSC-2G nuclear extracts (Fig. 3 B, arrowhead). By contrast, anti-Sp3 antibodies showed a relatively modest effect on both slowly and faster migrating complexes. On the other hand, in the case of primary culture of hepatocytes, anti-Sp3 antibodies completely diminished the intensity of the fast migrating Box 3A-bound complex (Fig. 3 B, arrowhead), whereas anti-Sp1 antibodies had a modest influence on both slowly and faster migrating complexes. Neither anti-Sp2 nor anti-Sp4 antibodies interfered with the complex formation with Box 3A probe (Fig. 3 B). To analyze semi-quantitatively the amounts of Sp1 and Sp3 proteins present in HSC and parenchymal hepatocytes, nuclear proteins prepared from both cell types were subjected to Western blot analyses using anti-Sp1 or anti-Sp3 antibodies. The results indicated that a larger amount of Sp1 protein was detected in nuclear extracts prepared from freshly isolated HSC than in those from hepatocytes (Fig.3 C). Conversely, more Sp3 protein was present in nuclei from parenchymal hepatocytes than those from HSC (Fig. 3 C). Likewise, nuclear extracts prepared from CFSC-2G cells contained a larger amount of Sp1 and less Sp3 protein as compared with hepatocyte nuclei (data not shown). The gel mobility shift assays and Western blot analyses of nuclear proteins indicated a difference in the relative amounts of Box 3A-bound Sp1 and Sp3 between CFSC-2G cells and primary culture of hepatocytes. Thus, we next examined the effects of overexpression of Sp1 or Sp3 onCOL1A2 transcription in both cell types. Transfection of CFSC-2G cells with an Sp1 expression plasmid significantly increased basal transcription of the −378COL1A2·LUC construct (Fig.4). TGF-β treatment further increasedCOL1A2 transcription in Sp1-transfected cells. Interestingly, overexpression of Sp1 in primary culture of hepatocytes not only increased basal COL1A2 transcription but also conferred TGF-β responsiveness to the cells (Fig. 4). In contrast to these stimulatory effects of Sp1 overexpression, transfection of CFSC-2G cells with an Sp3 expression plasmid did not affect the basal transcription levels (Fig. 4). More importantly, overexpression of Sp3 in CFSC-2G cells completely abolished TGF-β-elicited COL1A2 stimulation (Fig. 4). We next examined the effects of co-transfecting either Sp1 or Sp3 expression vector together with an expression plasmid encoding Smad3, an intracellular mediator of TGF-β signal transduction. Overexpression of Smad3 in CFSC-2G cells significantly increased transcription of the −378COL1A2·LUC construct (Fig.5 A). Whereas co-transfection with an Sp1 expression plasmid resulted in a further increase in −378COL1A2·LUC transcription, overexpression of Sp3 did not affect Smad3-stimulated COL1A2 transcription (Fig.5 A). Because there are at least two more Sp1 binding sites downstream of the TbRE (Fig. 1; Refs. 29Hasegawa T. Zhou X. Garrett L.A. Ruteshouser E.C. Maity S.N. de Crombrugghe B. Nucleic Acids Res. 1996; 24: 3253-3260Crossref PubMed Google Scholar, 30Ihn H. Ohnishi K. Tamaki T. LeRoy E.C. Trojanowska M. J. Biol. Chem. 1996; 271: 26717-26723Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar), we also performed co-transfection experiments using the −378COL1A2·LUC construct containing an internal deletion of the −183 to −108 segment (−378COL1A2del·LUC in Fig. 1). Basal transcriptional activity of this internally deleted construct was approximately one-fourth of that of the parental −378COL1A2·LUC construct (Fig. 5 A), indicating that the two Sp1 binding sites present in the −183 to −108 segment are also important for basal COL1A2 transcription. However, co-transfection of Smad3 and Sp1 expression plasmids resulted in the same level (∼3-fold) of transcriptional activation with both −378COL1A2·LUC and −378COL1A2del·LUC constructs (Fig.5 A). It is therefore suggested that Smad3 might interact with the TbRE-bound Sp1, rather than with the downstream Sp1, to mediate TGF-β-elicited COL1A2 stimulation. Functional interaction between Sp1 and Smad3, but not between Sp3 and Smad3, was further confirmed using the bacterial GAL4 system. Transfection of CFSC-2G cells with an expression plasmid encoding GAL4 DNA binding domain-fused Sp1 protein (pMSp1) increased transcription of the pG5-luc reporter gene ∼18-fold. Co-transfection with a Smad3 expression plasmid further increased Sp1-stimulated transcription, although Smad3 by itself had no effect on transcription (Fig.5 B). By contrast, transfection with an expression plasmid encoding a transactivation domain-deleted Sp1 (pMDNSp1) resulted in less than 4-fold increase in pG5-luc transcription, and it was not further stimulated by co-transfection with a Smad3 expression plasmid (Fig. 5 B). On the other hand, overexpression of an Sp3·GAL4 fusion protein (pMSp3) stimulated transcription 8-fold, which was not affected by co-transfection with a Smad3 expression plasmid (Fig. 5 B). In the last set of experiments, we examined physical interactions between Sp1 and Smad3 and between Sp3 and Smad3 using immunoprecipitation followed by Western blot analysis. We first attempted to examine interactions between endogenous GC box binding factors and Smad3 using CFSC-2G cells, but failed to detect any co-immunoprecipitated proteins (data not shown). It was not clear whether this was because of a lack of interaction between Sp1/Sp3 and Smad3 or because of relatively small amounts of these proteins present in CFSC-2G cells. We therefore transfected COS-7 cells with a Myc-tagged Smad3 expression plasmid together with either Sp1 or Sp3 expression vector, in the presence or absence of HA-tagged constitutive active TGF-β type I receptor (ALK5TD). Direct immunoblotting of whole cell lysates with anti-Myc, anti-Sp1, or anti-Sp3 antibodies confirmed expression of Myc-tagged Smad3, Sp1, and Sp3, respectively, in transfected COS-7 cells (Fig.6). In the absence of ALK5TD, anti-Sp1 antibodies hardly detected immunocomplexes, which had been first precipitated with anti-Myc antibodies recognizing Myc-tagged Smad3. By contrast, overexpression of ALK5TD markedly enhanced the physical interaction between Sp1 and Smad3 (Fig. 6), indicating that the interaction is TGF-β-dependent. On the other hand, immunoblotting using anti-Sp3 antibodies failed to detect Smad3-bound Sp3, either in the presence or absence of ALK5TD (Fig. 6). In this study, we demonstrated that different molecular mechanisms control COL1A2 transcription in activated HSC and parenchymal hepatocytes. Our experimental data indicated the differential roles of Sp1 and Sp3 in COL1A2 regulation. They also suggested that, in parenchymal hepatocytes, predominant binding of Sp3 to the Box 3A sequence and a lack of interaction with Smad3 may account, at least in part, for relatively low levels ofCOL1A2 transcription and loss of TGF-β responsiveness. It should be noted that, although we confirmed more than 95% purity of hepatocytes, there still remained a small number of HSC contaminated in the culture (31Maher J.J. Bissell D.M. Friedman S.L. Roll F.J. J. Clin. Invest. 1988; 82: 450-459Crossref PubMed Scopus (131) Google Scholar), which may respond to TGF-β. Despite possible underestimate of the results because of this contamination of HSC, cell transfection assays clearly indicated that the COL1A2promoter containing the TbRE did not show TGF-β responsiveness when transfected into primary culture of hepatocytes. Sp1 and Sp3 are closely related proteins with very similar structural features (32Hagen G. Muller S. Beato M. Suske G. Nucleic Acids Res. 1992; 20: 5519-5525Crossref PubMed Scopus (526) Google Scholar). They bind to the common GC-rich sequence named the GC box with the same specificity and affinity and regulate gene transcription (33Hagen G. Muller S. Beato M. Suske G. EMBO J. 1994; 13: 3843-3851Crossref PubMed Scopus (657) Google Scholar). However, Sp3 often acts as a transcriptional repressor by competitively binding to the Sp1-bound GC box sequences (33Hagen G. Muller S. Beato M. Suske G. EMBO J. 1994; 13: 3843-3851Crossref PubMed Scopus (657) Google Scholar) and/or by expressing internally initiated Sp3 proteins functioning as potent inhibitors of Sp1/Sp3-mediated transcription (34Kennett S.B. Udvadia A.J. Horowitz J.M. Nucleic Acids Res. 1997; 25: 3110-3117Crossref PubMed Scopus (233) Google Scholar). It is now recognized that Sp3 can either activate or repress transcription of target genes depending on the cell type, the context of DNA binding sites, and the interactions with other nuclear factors (35Majello B. De Luca P. Lania L. J. Biol. Chem. 1997; 272: 4021-4026Abstract Full Text Full Text PDF PubMed Scopus (257) Google Scholar). It has been previously reported that both Sp1 and Sp3 stimulateCOL1A2 transcription when transfected intoDrosophila Schneider cells lacking endogenous Sp transcription factors (36Ihn H. Trojanowska M. Nucleic Acids Res. 1997; 25: 3712-3717Crossref PubMed Scopus (99) Google Scholar). By contrast, it has been shown by others that co-transfection of an Sp3 expression plasmid inhibits Sp1-stimulated COL1A1 transcription in the same insect cells (37Chen S.-J. Artlett C.M. Jimenez S.A. Varga J. Gene. 1998; 215: 101-110Crossref PubMed Scopus (70) Google Scholar). However, neither of the studies examined the effects of Sp1 or Sp3 on TGF-β-elicited COL1A2 stimulation. The present study clearly demonstrated that, although Sp3 functioned as a weaktrans-activator of transcription in the bacterial GAL4 system, it did not increase basal levels of COL1A2transcription when transfected into CFSC-2G cells. More importantly, overexpression of Sp3 abolished TGF-β responsiveness in CFSC-2G cells. Different behaviors of Sp3 in regulating COL1A2transcription in Drosophila Schneider cells and mammalian CFSC-2G cells might be attributed to the presence of interacting factor(s) in the latter cells (35Majello B. De Luca P. Lania L. J. Biol. Chem. 1997; 272: 4021-4026Abstract Full Text Full Text PDF PubMed Scopus (257) Google Scholar). On the other hand, overexpression of Sp1 in primary culture of hepatocytes conferred TGF-β responsiveness with regard to COL1A2 transcription. A family of proteins termed Smad have been identified (38Heldin C.-H. Miyazono K. ten Dijke P. Nature. 1997; 390: 465-471Crossref PubMed Scopus (3366) Google Scholar) and found to play important roles in the intracellular signal transduction pathways of the TGF-β superfamily members (22Murata Y. Kim H.G. Rogers K.T. Udvadia A.J. Horowitz J.M. J. Biol. Chem. 1994; 269: 20674-20681Abstract Full Text PDF PubMed Google Scholar). Some of them, Smad3 and Smad4, have been shown to bind to the so-called CAGACA sequence present in the promoters of several TGF-β-inducible genes including plasminogen activator inhibitor-1 (39Dennler S. Itoh S. Vivien D. ten Dijke P. Huet S. Gauthier J.-M. EMBO J. 1998; 17: 3091-3100Crossref PubMed Scopus (1596) Google Scholar), junB (40Jonk L.J.C. Itoh S. Heldin C.-H. ten Dijke P. Kruijer W. J. Biol. Chem. 1998; 273: 21145-21152Abstract Full Text Full Text PDF PubMed Scopus (513) Google Scholar), p21WAF1/Cip1 (41Moustakas A. Kardassis D. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 6733-6738Crossref PubMed Scopus (322) Google Scholar), and Smad7 (42Nagarajan R.P. Zhang J. Li W. Chen Y. J. Biol. Chem. 1999; 274: 33412-33418Abstract Full Text Full Text PDF PubMed Scopus (207) Google Scholar). Box B of the COL1A2 promoter also contains a CAGACA sequence (−265 to −260) (39Dennler S. Itoh S. Vivien D. ten Dijke P. Huet S. Gauthier J.-M. EMBO J. 1998; 17: 3091-3100Crossref PubMed Scopus (1596) Google Scholar, 40Jonk L.J.C. Itoh S. Heldin C.-H. ten Dijke P. Kruijer W. J. Biol. Chem. 1998; 273: 21145-21152Abstract Full Text Full Text PDF PubMed Scopus (513) Google Scholar), and it has been shown that Smad3 binds to this sequence in vitro and stimulates COL1A2transcription (12Chen S.-J. Yuan W. Mori Y. Levenson A. Trojanowska M. Varga J. J. Invest. Dermatol. 1999; 112: 49-57Abstract Full Text Full Text PDF PubMed Scopus (377) Google Scholar, 13Chen S.-J. Yuan W. Lo S. Trojanowska M. Varga J. J. Cell. Physiol. 2000; 183: 381-392Crossref PubMed Scopus (171) Google Scholar). We have recently revealed that an interaction between Sp1 and Smad3 is critical in mediating the stimulatory effect of TGF-β onCOL1A2 transcription in NIH 3T3 fibroblasts (14Zhang W. Oul J. Inagaki Y. Greenwel P. Ramirez F. J. Biol. Chem. 2000; 275: 39237-39245Abstract Full Text Full Text PDF PubMed Scopus (214) Google Scholar). We demonstrated a functional interaction between Sp1 and Smad3/Smad4 using co-transfection experiments. In addition, it was found that, despite the presence of CAGACA sequence in Box B, recombinant Smad3 and Smad4 proteins did not bind to the −313 to −255 sequence by themselves. Nor could they stimulate COL1A2 transcription in fibroblasts stably transfected with an expression plasmid encoding the dominant-negative form of Sp1 (14Zhang W. Oul J. Inagaki Y. Greenwel P. Ramirez F. J. Biol. Chem. 2000; 275: 39237-39245Abstract Full Text Full Text PDF PubMed Scopus (214) Google Scholar). Only in the presence of Sp1 protein, Smad3 and Smad4 were able to bind to this sequence and synergistically stimulate COL1A2 transcription. These data suggest the possibility of physical interactions between Sp1 and Smad3/Smad4 (14Zhang W. Oul J. Inagaki Y. Greenwel P. Ramirez F. J. Biol. Chem. 2000; 275: 39237-39245Abstract Full Text Full Text PDF PubMed Scopus (214) Google Scholar). The present study further confirmed these functional and physical interactions between Sp1 and Smad protein(s) in activated HSC by utilizing the bacterial GAL4 system and co-immunoprecipitation experiments. It also revealed that similar interactions were not observed between Sp3 and Smad3. A recent study has revealed that the glutamine-rich transactivation domain of Sp1 and the MH1 domain of Smad3 mediate the interaction between these two proteins bound to the p21WAF1/Cip1 gene promoter, which is stimulated by TGF-β (43Pardali K. Kurisaki A. Moren A. ten Dijke P. Kardassis D. Moustakas A. J. Biol. Chem. 2000; 275: 29244-29256Abstract Full Text Full Text PDF PubMed Scopus (345) Google Scholar). Consistent with these results, the present study showed that co-transfection of an expression plasmid encoding transactivation domain-deleted Sp1 together with a Smad3 expression vector did not increase transcription in the bacterial GAL4 system. It has been previously reported that the glutamine-rich transactivation domain of Sp1 cannot be replaced by the homologous sequence of Sp3 (33Hagen G. Muller S. Beato M. Suske G. EMBO J. 1994; 13: 3843-3851Crossref PubMed Scopus (657) Google Scholar). Taken together, it is conceivable that a lack of interaction between Sp3 and Smad3 may account, at least in part, for the loss of TGF-β responsiveness of COL1A2 transcription in parenchymal hepatocytes. It has been shown that Smad2, another TGF-β responsive Smad, is also capable of interacting with Sp1 (43Pardali K. Kurisaki A. Moren A. ten Dijke P. Kardassis D. Moustakas A. J. Biol. Chem. 2000; 275: 29244-29256Abstract Full Text Full Text PDF PubMed Scopus (345) Google Scholar, 44Feng X.-H. Lin X. Derynck R. EMBO J. 2000; 19: 5178-5193Crossref PubMed Scopus (352) Google Scholar). Our recent study revealed that Smad2 did not bind to the TbRE of the COL1A2 promoter (14Zhang W. Oul J. Inagaki Y. Greenwel P. Ramirez F. J. Biol. Chem. 2000; 275: 39237-39245Abstract Full Text Full Text PDF PubMed Scopus (214) Google Scholar). Nor did transfection of NIH 3T3 cells with a Smad2 expression plasmid stimulate basal COL1A2 transcription (14Zhang W. Oul J. Inagaki Y. Greenwel P. Ramirez F. J. Biol. Chem. 2000; 275: 39237-39245Abstract Full Text Full Text PDF PubMed Scopus (214) Google Scholar). On the other hand, under a certain experimental condition, overexpression of Smad2 in primary culture of skin fibroblasts markedly increased TGF-β responsiveness of COL1A2 transcription without affecting basal transcription level (45Inagaki Y. Mamura M. Kanamaru Y. Greenwel P. Nemoto T. Takehara K. ten Dijke P. Nakao A. J. Cell. Physiol. 2001; 187: 117-123Crossref PubMed Scopus (114) Google Scholar). Functional roles of Smad2 in TGF-β-stimulated COL1A2 transcription have not been fully understood, and experiments are in progress to clarify possible contribution of Smad2 to cell lineage-specificCOL1A2 transcription. In conclusion, the present study is the first to demonstrate, at the molecular level, differences in cell lineage-specific regulation ofCOL1A2 transcription in activated HSC and parenchymal hepatocytes. It illustrates that two members of the GC box binding transcription factor family participate in regulation ofCOL1A2 transcription through differential interaction with Smad3. These results lead to not only better understanding of regulatory mechanisms responsible for cell type-specific gene expression but also the development of novel therapeutic means for fibrotic diseases in various organs by suppressing pathologically activated collagen gene transcription. We thank Drs. Rik Derynck, Gregory A. Elder, and Jonathan M. Horowitz for providing us with valuable expression plasmids and to Dr. Yoshikazu Sowa for his generous gift of GAL4 binding domain fusion constructs as well as useful discussion. We also thank Drs. Scott L. Friedman, Francesco Ramirez, Marcos Rojkind, and Shizuko Tanaka for their helpful advice and critical suggestions.

Referência(s)