Regulation of Clusterin Gene Expression by Transforming Growth Factor β*

Transforming growth factor β (TGFβ) induces the expression of a wide variety of genes in many cell types. Our previous studies have shown that TGFβ stimulates both clusterin mRNA and protein levels, and induces its accumulation in the nucleus of CCL64 cells. To further investigate the molecular mechanism of clusterin mRNA induction by TGFβ, we created a 1.3-kilobase rat clusterin promoter/luciferase reporter construct. We demonstrate that TGFβ enhances luciferase activity 2.5–6-fold in transient transfection assays of epithelial, endothelial, and fibroblast cell lines. Deletional analysis reveals that an AP-1-binding site (5′-TGAGTCA) in the minimal promoter region is necessary for initiating transactivation by TGFβ. A single T to G base mutation in the AP-1 site (5′-TGAGGCA) abolishes TGFβ-induced clusterin promoter transactivation. In transcription factor decoy experiments, 23-mer oligonucleotides of wild type AP-1 reduce TGFβ induction of clusterin mRNA levels and promoter transactivation, while an oligonucleotide containing the mutated AP-1 site has no effect. Two specific protein kinase C inhibitors, GF109203X and calphostin C, block TGFβ-induced clusterin mRNA levels and promoter transactivation. Together these results indicate that TGFβ regulates clusterin gene expression through an AP-1 site and its cognate transcription factor AP-1, and requires the involvement of protein kinase C.

The transforming growth factor-␤ (TGF␤) 1 family of cytokines consists of multifunctional proteins which play important roles in regulating cell growth, development, and differentiation (1)(2)(3)(4)(5)(6). A number of structural and metabolic proteins, such as fibronectin and its receptor, collagen, collagenase, plasminogen activator inhibitor type-1, and clusterin, have been shown to be regulated by TGF␤. In addition, expression of some cellular oncogenes, such as c-jun, junB, and c-fos, and TGF␤ itself are also regulated by TGF␤ (2,5,(7)(8)(9)(10)(11). TGF␤ regulates ex-pression of its responsive genes through binding to specific membrane TGF␤ receptors, which possess Ser/Thr kinase activity, and triggering an unknown signaling cascade to modulate the interaction of transcription factors and their cognate cis-elements (2,(12)(13)(14). Recent studies have shown that Smad proteins, which are postulated to function as TGF␤ receptorregulated transcription factors, may act as cellular mediators in TGF␤ signaling of mammalian cells and play a critical role in transmitting the TGF␤ signal to the nucleus (15,16). Protein kinase C and other protein kinases have also been implicated in TGF␤-mediated regulation of gene expression. These kinases may participate in recruitment of transcription factors, such as activator protein 1 (AP-1), to modulate TGF␤ responsive gene expression (6,7,15,(17)(18)(19)(20)(21)(22). Many TGF␤ responsive genes, such as plasminogen activator inhibitor type-1, contain AP-1 consensus sequences in their regulatory region, and the sequence is required for TGF␤ regulation of genes in both growth-stimulated and growth-inhibited cell lines (7,8,(23)(24)(25)(26). Sp1 has been shown to participate in the regulation of human ␣2(I)-collagen and p21/Waf1/Cip1 gene expression by TGF␤ (27,28). Nuclear factor 1 also appears to be involved in expression of several genes regulated by TGF␤ (29). TGF␤ modulates interaction of theses transcription factors and their cognate elements in a coordinated manner to specifically regulate TGF␤-responsive gene expression. However, the signaling pathway(s) through which TGF␤ modulates gene responses in mammalian cells remains largely unknown.
The clusterin protein was first discovered in ram rete testis fluid as an ϳ80-kDa heterodimeric glycoprotein that facilitated the aggregation of a variety of cells in culture (30). A number of homologues of clusterin have been identified in several species (31). Clusterin is present in almost all mammalian body fluids and can also be induced or constitutively expressed in almost all cell types (30,31). The protein has been implicated in a variety of biological processes including lipid transport, inhibition of complement attack, sperm maturation, epithelial cell differentiation, and membrane remodeling during apoptosis and implantation (31)(32)(33)(34)(35)(36)(37). Analysis of the 5Ј-regulatory region of the clusterin gene has revealed TGF␤ inhibitory elements as well as AP-1, Sp1, and AP-2 regulatory elements in the quail, rat, and human clusterin genes (38 -41). These elements are postulated to be responsible for the modulation of clusterin gene expression observed during cell differentiation, development, and embryogenesis (31, 35, 39 -41). It has also been demonstrated that clusterin gene expression can be regulated by TGF␤ in a cell type-dependent manner (35,42). For example, TGF␤ down-regulates clusterin mRNA levels in porcine smooth muscle cells (35), induces its gene expression in rat astrocytes in the presence of oligodendrocytes and microglia, while repressing its message in monotypic cultures of astrocytes (42). However, the mechanism(s) of regulation of clusterin gene expression by TGF␤ is unknown.
We have previously reported that TGF␤ induces expression of clusterin protein and rapidly stimulates clusterin mRNA levels in a mink lung epithelial cell line CCL64 cells (Mv1Lu). We have also demonstrated that TGF␤ selectively induces accumulation of clusterin protein in the nucleus (11,43). To further dissect the mechanism of TGF␤-mediated regulation of clusterin gene expression, we have analyzed the intracellular signaling pathway through which TGF␤ modulates gene expression. Our results demonstrate that TGF␤ regulates clusterin gene expression in epithelial, fibroblasts, and in a primary culture of bovine aortic endothelial cells, at least in part, at the transcriptional level. An AP-1 site in the rat clusterin 5Ј-promoter region is responsible for the regulation of clusterin gene expression by TGF␤, and the effect of TGF␤ on clusterin gene expression requires the involvement of protein kinase C.
Northern Blot Analysis-Cells from confluent cultures were harvested and seeded at 3ϳ5 ϫ 10 6 cells per 100-mm plate. After attaining 50% confluence the plates were treated with TGF␤ (5 ng/ml) for different periods. For Northern blot analysis with serum-starved CCL64 cells, the serum medium was replaced with DMEM/F-12 medium containing 0.1% bovine serum albumin once the cells had reached 50ϳ60% confluence. The cells were incubated at 37°C for 24 h prior to TGF␤ stimulation. Total RNA was extracted by guanidine isothiocyanate according to Chomczynski and Sacchi (45), and separated (20 g of RNA/well) in 1.2% formaldehyde gels. The RNA was transferred to NytranPlus nylon membrane (Schleicher & Schuell) by capillary blotting and hybridized with a random-labeled (Pharmacia) full-length human clusterin cDNA at 2-4 ϫ 10 6 cpm/ml. The same blot was stripped with 55% formamide, 2 ϫ SSC, 1% SDS for 60 min at 65°C, and the membrane was re-probed with 32 P-labeled human cyclophilin cDNA (1B15) to ensure equal loading of the RNA. The hybridization strength was analyzed by autoradiography (11).
DNA Subcloning and Deletions-pLLTRPM2 ϩ plasmid (kindly provided by Dr. Martin Tenniswood) containing a rat clusterin 5Ј-regulatory region from ϩ57 to Ϫ1297 was digested with SphI and HindIII and subcloned into pGL2-basic luciferase reporter vector (Promega) to form a pRAL plasmid for transient transfection, deletion, and mutagenesis experiments. Four deletions were obtained using restriction enzymes: pRAL-S/A starting from Ϫ218 to ϩ57 was generated by digestion of pRAL with SacI and ApaI; pRAL-M/Nsi deletion contains promoter region from Ϫ729 to ϩ57 produced by restriction enzymes MluI and NsiI. The others are internal deletions pRAL-N/A (Ϫ728 to Ϫ218 was deleted) and pRAL-E/E (Ϫ609 to Ϫ30 was deleted) generated by NsiI/ ApaI and Eco47III, respectively. To make deletions around the AP-1 site, the following series of primers were synthesized according to the sequence of pRAL and used for polymerase chain reaction amplification and subcloning (see Fig. 2B): P1, 5Ј-GTTTGCAGCCAGCCAAAG-3Ј; P2, 5Ј-CCAGAGGAATTCATTATCAG-3Ј; P3, 5Ј-AGAATGCCGGGGAATG-CACTAGGAG-3Ј; P4, 5Ј-CAGAAAGCTCCTAGTGCATTC-3Ј.
Site-directed Mutagenesis-A single site-directed mutant in the AP-1-binding site of the promoter was created basically as described by Deng and Nickoloff (46). The mutagenic primer, 5Ј-CTGGCGTGAG-GCACGCAGGTTTG-3Ј, corresponds to Ϫ62 to Ϫ85 of the promoter region centered by a mutated AP-1-binding site (the underlined represents the base that was mutated). The selection primer, 5Ј-GCGACT-GGTGAGGCCTCAACCGGCTTC-3Ј, contains a single base mutation in a ScaI restriction site within the vector; therefore, the mutant will not be cleaved by ScaI. To create the mutation, the mutagenic and selection primers were phosphorylated with T 4 kinase (Promega) and annealed to denatured plasmid pRAL to generate mutated first strand DNA by Klenow fragment. Then the DNA was cut by ScaI to digest wild type plasmid while the DNA hybrid remained intact. Competent BMH71-18 mutS cells (CLONTECH) were transformed by the DNA hybrid and incubated at 37°C overnight to amplify mutated plasmid. After incubation, the plasmid was isolated and subjected to secondary ScaI digestion. The undigested, mutated plasmid was used to transform JM109 cells (Promega) and colonies were picked for plasmid preparation. The mutated plasmid was confirmed by sequencing and termed pT2G. The wild type rat clusterin 5Ј-promoter region contains an AP-1 consensus sequence as 5Ј-TGAGTCA-3Ј, the mutated AP-1 site has the sequence 5Ј-TGAGGCA-3Ј at the same position in the promoter region.
Transient Transfection Assays-Transfection assays were carried out using liposomes prepared as described by Campbell (47). Briefly, 50 l of lipid mixture containing 13.4 mM 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (Avanti Lipid Inc.) and 6.6 mM dimethyldiocatadecyl ammonium bromide (Sigma) in 100% ethanol were added to 1 ml of sterile water with vortexing. For transfection, cells were plated in 6-well plates at a density of 2 ϫ 10 5 cells/well and incubated in DMEM/ F-12 medium containing 10% newborn calf serum at 37°C until 70 -80% confluent. For transfection assay of CCL64 cells, 30 l of liposome reagent (ϳ3 g of lipid) was added to 1 ml of serum-free medium, mixed with 2 g of luciferase reporter plasmid and 1 g of pSV-␤-galactosidase plasmid (Promega) as an internal control, and incubated at room temperature for 30 min to form a liposome-DNA complex. Prior to transfection, cells were washed twice with serum-free DMEM/F-12 medium, then 1 ml of liposome-DNA complex was added to the cells and incubated at 37°C for 5 h before adding TGF␤. For transfection assays with other cell lines, 20 l of liposome reagent, 1 g of reporter plasmid DNA, and 0.2 g of pRL-CMV plasmid (Promega) as an internal control were used. All other procedures were the same as for transfections of CCL64 cells. After removing the transfection medium, serum or serumfree medium containing TGF␤ was added to the cells and incubated for 16 or 20 h at 37°C prior to extraction of cellular lysates.
Decoy Experiments of Transcription Factors-Sense and antisense 23-mer oligonucleotides, containing either AP-1-binding site (AP-1) or its mutant (T2G) from the rat clusterin promoter region, were synthesized and annealed to form double-stranded DNA. The sequence of the 23-mer AP-1 oligonucleotide is 5Ј-CTGGCGTGAGTCACGCAG-GTTTG-3Ј (the AP-1 consensus site is bold faced), and that of the T2G oligonucleotide is 5Ј-CTGGCGTGAGGCACGCAGGTTTG-3Ј (the T to G single site mutation is underlined). For decoy assays analyzed by Northern blot, 20 nmol of DNA was mixed with 30 l of liposome reagent in 2 ml of serum-free medium, and incubated at room temperature for 30 min to form a liposome-DNA complex. Cells, 80% confluent, were washed with serum-free medium prior to the addition of the liposome-DNA complex. After incubation for 2 h, the cells were stimulated with 5 ng/ml TGF␤ for 10 h and subjected to RNA extraction and Northern blot analysis. To perform decoy experiments in transient transfection assays, 0.7 g of oligonucleotide was added along with pRAL (100:1 molar ratio) into liposome reagent for transfection and luciferase activity.
Luciferase and Galactosidase Activity Assays-Following transient transfection and TGF␤ stimulation, cells were washed with phosphatebuffered saline two times and lysed at room temperature in 300 l of lysis buffer (25 mM Tris phosphate, pH 7.8, 2 mM dithiothreitol, 2 mM CDTA, 10% glycerol, 1% Triton X-100). The cell lysate was scraped into a microcentrifuge tube and centrifuged at high speed for 30 s at room temperature to remove cell debris. Cell lysates (10 l) were mixed with 100 l of luciferase substrate buffer (Promega) and luciferase activity was measured immediately (ML2250 Microtiter Plate Luminometer, Dynatech Laboratories). For lysates from cells transfected with pRL-CMV plasmid as an internal control construct, the reporter plasmid and control plasmid luciferase activities were measured with Promega's Dual-Luciferase Reporter Assay System according to manufacture's protocol. To measure galactosidase activity, 150 l of 2 ϫ ␤-galactosidase buffer (0.2 M NaPO 4 , pH 7.3, 2 mM MgCl 2 , 0.1 M ␤-mercaptoethanol, 1.33 mg/ml o-nitrophyenyl ␤-D-glactopyranoside) was added to 150 l of cell lysate. Following incubation at 37°C overnight, 600 l of H 2 O was added to ␤-galactosidase reaction mixture and A 420 was read with a spectrophotometer (UV160U, Shimadzu) to quantify ␤-galactosidase activity. The promoter/reporter luciferase activity was normalized to ␤-galactosidase activity or renilla luciferase activity of the Dual-Luciferase Reporter Assay System, respectively.

RESULTS
TGF␤ Induces Clusterin mRNA Synthesis-We have previously demonstrated that TGF␤ induces clusterin protein synthesis and accumulation of clusterin in the nucleus of CCL64 cells (11). However, the TGF␤ signaling pathway mediating this induction remains unknown. Therefore, we were interested in studying the regulation of clusterin gene expression by TGF␤ in an attempt to identify some of the signaling components involved. Northern blot analysis showed that TGF␤ in-duces clusterin mRNA as early as 30 min after TGF␤ treatment and continues about 24 h in rapidly growing, asynchronous CCL64 cells (Fig. 1, A and B). To determine whether the induction of clusterin RNA synthesis was a consequence of the growth-inhibitory effect of TGF␤ on the cells, CCL64 cells were serum starved in serum-free DMEM/F-12 medium for 24 h, prior to treatment with TGF␤ for different periods. As shown in Fig. 1C, TGF␤ also induces clusterin mRNA synthesis in serum-starved, quiescent cells, indicating that induction of clusterin mRNA level by TGF␤ is not a result of growth inhibition of the cells. The induction of clusterin mRNA by TGF␤ is not restricted to mink lung epithelial cells. The data presented in Fig. 1D demonstrates that TGF␤ can also induce clusterin gene expression in other cell types. In two fibroblasts cell lines, 10T1/2 and 3TP, two epithelial cell lines, CCL64 and HeLa, and in primary BAEC, TGF␤ induces clusterin mRNA levels following a 16-h treatment.
To determine whether the induction of clusterin mRNA by TGF␤ is at the transcriptional level, we performed transient transfection assays in these different cell lines with a luciferase reporter gene (pRAL) containing the 1.3-kilobase 5Ј-regulatory region of rat clusterin gene (Fig. 2B). Following transfection, cells were treated with 5 ng/ml TGF␤ for 16 or 20 h. The data ( Fig. 2A and Fig. 4) show that TGF␤ induces promoter transactivation with a 2.5-6-fold elevation in luciferase activity in the five cell types analyzed. In transient transfection assays of CCL64 cells, we used this same reporter construct lacking the TATA box in the regulatory region as a negative control ( Fig.  2A). This construct failed to induce luciferase activity in response to TGF␤. The results provide supportive evidence that TGF␤ induces clusterin gene expression, at least in part, at the transcriptional level.
Deletional Analysis of a Rat Clusterin Gene 5Ј-Promoter Region-To analyze the TGF␤-responsive cis-element(s) of the clusterin promoter, we constructed a series of 5Ј-deletions of the promoter region of rat clusterin gene. Since the 1.3-kilobase promoter region contains Sp1, AP-2, and AP-1 consensus sites, we decided to make deletions containing different combinations of these conventional binding sites as the first step in our promoter analysis (Figs. 2 and 3). The pRAL-M/Nsi deletion with 586 base pairs truncated from the 5Ј-promoter region, which still retains all three binding sites, is able to induce TGF␤ transactivation. Two other deletions, pRAL-N/A and pRAL-p2p4, which lack Sp1 and both Sp1 and AP-2 sites, respectively, still show TGF␤ induction of promoter transactivation similar to the full-length promoter/reporter construct. This suggests that Sp1 and AP-2 are not necessary for clusterin transactivation by TGF␤. However, deletions pRAL-E/E, pRAL-p1p2, and pRAL-AP(Ϫ), which remove the AP-1-binding site in the promoter region, abolished TGF␤-induced promoter transactivation (Fig. 3). The data indicate that the AP-1-binding site in the promoter region is essential for TGF␤-mediated regulation of clusterin gene expression.
The AP-1 Site Is Required for TGF␤-mediated Induction of Clusterin Gene Transcription-While the deletional analyses Luciferase activity was measured as relative light units with a luminometer and normalized by ␤-galactosidase activity of SV40 ␤-galactosidase. Induction of promoter transactivation was calculated by comparing normalized luciferase activity of transfected cells treated with or without TGF␤. The data represent duplicate experiments. B, construction of promoter-reporter plasmids for transient transfection analyses. The pRAL was constructed by subcloning a 1326-base pair 5Ј-regulatory region of rat clusterin gene into a luciferase reporter vector pGL-2 basic (Promega). A series of promoter deletions were made based on pRAL. Primers p1/p2 and p1/p3 were used to create AP-1 deletions, and p2/p4 was used to construct a minimal promoter containing the AP-1-binding site.
revealed that the AP-1-binding site in the clusterin promoter region is important for TGF␤-mediated induction of clusterin gene expression, we needed to determine whether the AP-1 site is the exclusive element in the clusterin promoter responsible for TGF␤-mediated gene regulation. To test this, we created a single base mutation within the AP-1-binding site of the promoter region. The wild-type AP-1-binding site is 5Ј-TGAGTCA-3Ј, while in the mutant construct, the thymidine was replaced by guanine to form an AP-1 mutant, 5Ј-TGAGGCA-3Ј termed pT2G (the mutated base is underlined) (Fig. 4). The mutated promoter/luciferase construct, pT2G, contains a single site mutation in the 1.3-kilobase pair rat promoter region. Transfection assays in the various cell types with the wild-type vector show induction of luciferase activity by TGF␤; however, the mutated AP-1 promoter/reporter construct fails to show TGF␤induced transcriptional activation (Fig. 4).
Transcription Factor Decoy Experiments in CCL64 Cells-We next wished to investigate whether the AP-1 site of the promoter region is specifically responsible for TGF␤-mediated induction of endogenous clusterin expression in CCL64 cells. To address this question, we designed a 23-mer oligonucleotide containing either the wild-type (AP-1) or mutated (T2G) AP-1-binding site (Fig. 5C). These 23-mer oligonucleotides were transfected into CCL64 cells and used as transcriptional factor decoys in experiments analyzing TGF␤-mediated regulation of clusterin mRNA levels (Fig. 5A) and promoter transactivation (Fig. 5B). As shown in Fig. 5A, introduction of the wild-type AP-1 oligonucleotide into CCL64 cells dramatically reduced clusterin mRNA level upon TGF␤ stimulation, while the T2G mutated AP-1 oligonucleotide had no effect. In addition, when the AP-1 or T2G oligonucleotides are transiently co-transfected in excess (100:1 nmol) with the fulllength pRAL reporter construct (Fig. 5B), only the wild-type AP-1 inhibits TGF␤-induced luciferase activity. These experiments indicate that TGF␤-mediated induction of endogenous clusterin also involves the AP-1-binding site.
Protein Kinase C Is Involved in Mediating TGF␤-induced Clusterin Gene Expression-Previous research has suggested that several protein kinases, including protein kinase C (PKC), may participate in mediating TGF␤ signaling although their functions and substrates remain to be elucidated (17). We were interested in investigating if the PKC signaling pathway might be involved in mediating TGF␤-mediated regulation of clusterin gene expression. To address this question, we used two specific pharmacological inhibitors of PKC, namely GF109203X and calphostin C (48,49), to block TGF␤-mediated regulation of clusterin gene expression. As shown in Fig. 6A, pretreatment of CCL64 cells with a specific PKC inhibitor, GF109203X (48), blocked the ability of TGF␤ to induce clusterin gene expression in a dose-dependent manner. Another

FIG. 3. Deletional analyses of rat clusterin promoter region in transient transfection assays of CCL64 cells.
Full-length and deletions of rat clusterin promoter reporter plasmids were transfected into CCL64 cells as described under "Materials and Methods." TGF␤-induced luciferase activity was measured as relative light units and normalized by ␤-glactosidase activity of cotransfected SV40 ␤-galactosidase reporter plasmid (Promega). Induction fold was calculated by comparing normalized luciferase activity of TGF␤-treated cells and untreated control. In each assay, duplicate transfections were performed. These results are from representative experiments. Consensus binding sites are labeled and designated by filled ovals.

FIG. 4. Mutational analysis of rat clusterin promoter region.
The sequence of the AP-1-binding site is shown at the top of the figure. A thymidine in the wild type AP-1-binding site of full-length promoter region was mutated to guanine as indicated by the underlined letter. The wild type and mutant promoter/reporter constructs were transfected into different cells. Luciferase activity of the promoter-reporter construct was measured and normalized by ␤-galactosidase activity of SV40 ␤-galactosidase plasmid (CCL64 cells) or renilla luciferase activity of pRL-CMV plasmid (other cells). Duplicate transfections were performed for each assay. specific PKC inhibitor, calphostin C (49), at concentrations of 50 and 100 nM, was also effective in inhibiting TGF␤-induced clusterin gene expression as determined by Northern blot analyses (Fig. 6B) and promoter driven luciferase activity in CCL64 cells (Fig. 6C). Calphostin C at 100 nM was also effective in inhibiting TGF␤-induced promoter activity in HeLa, 3TP, 10T1/2, and BAEC cells (data not shown). Down-regulation of PKC activity by treatment of the cells with 200 ng/ml phorbol myristate acetate (PMA) for 20 h inhibited TGF␤-induced clusterin mRNA synthesis (data not shown). However, stimulation of PKC in CCL64 cells with 10 -100 ng/ml PMA for 0.5-10 h had no effect on clusterin mRNA levels with or without TGF␤ stimulation (data not shown).

DISCUSSION
Clusterin is a multifunctional protein that has been implicated in homeostatic control of lipoprotein metabolism, tissue repair and remodeling, sperm maturation, inhibition of complement mediated cell lysis, and epithelial cell differentiation (30). Expression of clusterin is well regulated during development, cell differentiation, and tissue remodeling, and can be modulated by cytokines, such as TGF␤ (31,50,51). Our previous studies have shown that TGF␤ enhances clusterin protein synthesis and RNA levels, and induces the translocation of a cytosolic form of clusterin to the nucleus in CCL64 cells (11). These effects of TGF␤ on clusterin led us to investigate the mechanisms of regulation of clusterin gene expression by TGF␤. Our data demonstrate that TGF␤ rapidly induces clusterin mRNA levels in a variety of cell lines, including epithelial (CCL64 and HeLa), fibroblast (10T1/2 and 3TP), and BAEC. The induction of clusterin by TGF␤ in mink lung epithelial cells (CCL64) occurs in both rapidly growing, asynchronous cells and in serum-starved G 0 -arrested cells. This suggests that this induction is not cell cycle dependent since it occurs not only during quiescence but also throughout the cycle. These results also suggest that the induction of clusterin is not secondary to TGF␤-induced growth arrest because TGF␤ is capable of inducing clusterin in an already G 0 -arrested cell population, as well in 3TP cells which are not growth arrested by TGF␤ (44). Clusterin mRNA can be induced as early as 30 min after TGF␤ stimulation and remains elevated for at least 24 h in the presence of TGF␤. This induced expression pattern of clusterin mRNA is similar to that observed with other stimuli where mRNA levels peak within several hours and remain elevated for over 24 h following stimulation (31,50).
There is over 80% similarity within the first 150 base pairs of the 5Ј-upstream regulatory region between the human and rat clusterin genes (40). Two conventional cis-elements, Sp1 and AP-1, have been found in the regulatory region of clusterin genes from human, rat, quail, and mouse (39,40). Several TGF␤-inhibitory elements have also been identified in the first intron of the rat clusterin gene and in the upstream regulatory region of the avian clusterin gene (38,40,41). These cis-elements have been implicated in the regulation of clusterin gene expression by various stimuli (37)(38)(39)(40)(41). For example, the T64 gene of quail embryo fibroblasts, corresponding to clusterin in mammals, contains an AP-1-binding site located at position Ϫ25 to Ϫ19 of the single transcriptional start site. Promoter transactivation of the T64 gene has been shown to be significantly induced by an active v-src oncoprotein, and the v-src response requires the AP-1 site and a purine-rich element (39). There is, however, no evidence to indicate that these cis-elements in the promoter region of clusterin genes regulate TGF␤ induction.
In the present study, a 1.3-kilobase rat clusterin promoter region has been used for identifying cis-elements mediating TGF␤ signaling. This promoter contains AP-1, AP-2, and Sp1 consensus binding sites. Our deletional analyses demonstrate that the AP-2 and Sp1 consensus sites, located at position Ϫ124

FIG. 5. Decoy experiments of transcription factors with oligonucleotides containing either a wild-type or mutant AP-1-binding site.
A, effect of AP-1 or T2G mutant oligonucleotides on clusterin mRNA production. CCL64 cells (80% confluent) were transfected with 23-mer oligonucleotides containing the AP-1 consensus binding site or a T to G single site AP-1 mutant (T2G), respectively, and treated with TGF␤ for 10 h. Cells were subjected to RNA extraction and Northern hybridization. B, effect of co-transfection of the oligonucleotides on pRAL transgene luciferase activity. AP-1 or T2G-mutant 23-mer oligonucleotides (0.7 mg) were co-transfected along with pRAL and SV40 ␤-galactosidase into CCL64 cells. The promoter driven luciferase activity was normalized to ␤-glactosidase activity of the SV40 ␤-galactosidase plasmid, and the induction fold was calculated as described in the legend to Fig. 3. C, sequences of the 23-mer oligonucleotides containing either an AP-1-binding site (AP-1) or a T to G mutated AP-1 site (T2G). The position of the oligonucleotide in the promoter region is also shown in the figure. The bold faced sequence represents the consensus AP-1-binding site and the underlined indicates the base that was mutated. and Ϫ372 from the transcriptional start site, respectively, are not involved in promoter transactivation by TGF␤. However, removal of the AP-1 consensus site, located at position Ϫ73 to Ϫ79 relative to the transcription start site, abates TGF␤ induction of promoter transactivation. The data indicate that TGF␤ modulates clusterin gene expression via an AP-1-binding site, and that this AP-1 site is required and sufficient for promoter transactivation by TGF␤. The importance of the AP-1 site in the induction of the clusterin gene by TGF␤ is confirmed by our mutagenesis experiments of the AP-1 site in the clusterin promoter region. A single base pair mutation, T to G in the AP-1binding site, abolished TGF␤-promoter transactivation, indicating that the AP-1 site is necessary for TGF␤-induced clusterin expression. Decoy experiments with 23-mer AP-1 oligonucleotides also demonstrate that the AP-1 site is required for TGF␤ induction of the endogenous clusterin gene. Transfection of the wild-type AP-1 oligonucleotide into CCL64 cells markedly decreases TGF␤ induction of clusterin mRNA levels and promoter transactivation. Transfection of the T to G mutated AP-1 failed to serve as a decoy of the endogenous AP-1 site in the clusterin gene and had no effect on the ability of TGF␤ to induce endogenous clusterin gene expression. The data indicate that TGF␤ regulation of endogenous clusterin expression in CCL64 cells is also mediated via the AP-1-binding site and its cognate transcription factors.
In addition to clusterin, a large number of TGF␤ responsive genes, such as JE/MCP-1(25), c-jun (7, 51), plasminogen acti-vator inhibitor type-1 (8), and ␣2(I)-collagen (62), contain AP-1-binding sites in their 5Ј-regulatory region. Expression of these genes can be modulated by TGF␤ and requires involvement of the AP-1-binding site and its cognate transcription factor. The collagenase gene has also been shown to be regulated by TGF␤ in a cell type specific manner through an AP-1 site. In fibroblast cells, for example, TGF␤ inhibits collagenase gene expression through an AP-1 site presumably via an upregulation of junB (63). Overexpression of junB mimics the inhibitory effects of TGF␤ on collagenase expression. In contrast, TGF␤-induced collagenase expression in keratinocytes is preceded by a transient elevation of c-jun expression, which is suggested to be a ubiquitous inducer of collagenase gene expression (63). These results demonstrate that cell-specific induction of different AP-1 family members, with opposite transactivating properties, are responsible for the differences observed in TGF␤ regulation of collagenase gene expression. Taken together, these data establish that AP-1 is a mediator of TGF␤-induced gene transcription. However, the mechanism through which TGF␤ modulates AP-1 activity and the individual AP-1 family members which are induced or suppressed by TGF␤, in a cell-type and gene-specific fashion, remain to be determined.
We postulated that since the promoter region of the rat clusterin gene contains an AP-1-binding site, which is required for TGF␤ induction, that PKC activity might mediate TGF␤ induction. It is well established that activation of PKC by 12-O-tetradecanoylphorbol-13-acetate results in increased synthesis of c-Fos and c-Jun proteins which are required for the formation of the AP-1 transcriptional complex (AP-1) (53,54). Upon activation and translocation to the nucleus, the AP-1 protein binds to the 12-O-tetradecanoylphorbol-13-acetate responsive elements mediating activation of 12-O-tetradecanoylphorbol-13-acetate inducible genes (7,53,54). The 12-O-tetradecanoylphorbol-13-acetate responsive element has been identified as an AP-1-binding site, and the AP-1 site as well as its cognate transcription factors are directly related to phorbol ester, a PKC regulatory reagent (51,(55)(56)(57). It has been postulated that activated PKC leads to dephosphorylation of c-Jun protein at sites which negatively regulate DNA binding activity and thereby augments AP-1 activity (64). PKC has also been implicated in TGF␤ signaling. In A549 human lung carcinoma cells, TGF␤ has been shown to induce promoter transactivation of 3TP-Lux, a TGF␤-inducible artificial promoter construct which contains 3 12-O-tetradecanoylphorbol-13-acetate responsive elements (17). TGF␤ induction of this promoter construct can be inhibited by prior treatment of the cells with PKC inhibitors and down-regulation of PKC with PMA (17). In the present study we demonstrate that inhibition of PKC activity by two distinct types of PKC inhibitors blocked TGF␤-induced clusterin mRNA levels and promoter transactivation, thus providing further support for the role of AP-1 in TGF␤ regulation of clusterin gene expression. Stimulation of PKC activity by treatment of the cells with PMA, however, had no effect on clusterin mRNA levels or TGF␤-induced clusterin mRNA levels. These results suggest that activation of PKC alone is not sufficient for regulating clusterin gene expression but that PKC activity is necessary for TGF␤-induced clusterin gene expression. These data are in agreement with the observation that PMA treatment does not coordinate phosphorylation of nuclear proteins in CCL64 cells, whereas TGF␤ stimulates nuclear protein phosphorylation (20). These results are also similar to those observed in activated T cells where full activation of AP-1 requires both the calcium-and PKC-dependent pathways (60). Perhaps TGF␤-inducible clusterin gene expression requires multiple signaling pathways to fully activate FIG. 6. Effect of protein kinase C inhibitors on induction of clusterin gene expression by TGF␤. A, GF109236X inhibits TGF␤induced clusterin mRNA levels in CCL64 cells. Cells were pretreated with 5, 15, and 25 mM GF109236X for 15 min (61) and TGF␤ was added for 12 h prior to the isolation of total RNA and Northern blot analysis. B, effect of calphostin C on TGF␤-induced clusterin mRNA. CCL64 cells (80% confluent) were pretreated with 100 nM calphostin C for 15 min under room lights followed by stimulation with TGF␤ (5 ng/ml) for the indicated times prior to RNA extraction and Northern hybridization. C, effect of calphostin C pretreatment on clusterin promoter transcriptional activity induced by TGF␤. After transfection of CCL64 cells with pRAL and SV40 ␤-gal, calphostin C was added to the medium at final concentrations of 50 and 100 nM and incubated for 15 min under room lights. Cells were stimulated with TGF␤ (5 ng/ml) for 16 h prior to isolation of cell extracts and measurement of luciferase activity.
AP-1 and that inhibiting one of these signaling cascades (i.e. PKC) results in insufficient activation of AP-1 to induce proper transcriptional regulation.
Recent studies in TGF␤ signaling indicate that binding of TGF␤ to its dimeric receptor complex, composed of the type I and type II receptors, initiates a serine/threonine phosphorylation cascade that involves the Smad proteins. These signaling molecules are phosphorylated on serine and threonine residues by the type I TGF␤ receptor and once activated translocate to the nucleus where they are postulated to interact with other Smad proteins and/or transcription factors to initiate target gene expression (65). Future research will be directed at determining how these Smad proteins interact with members of the AP-1 family to regulate gene transcription and how PKC regulates this interaction.