Involvement of transcription factor Ets-1 in the expression of the α3 integrin subunit gene

Authors


Correspondence

T. Tsuji, Department of Microbiology, Hoshi University School of Pharmacy and Pharmaceutical Sciences, 2-4-41 Ebara, Shinagawa-ku, Tokyo 142-8501, Japan

Fax: +81 3 5498 5753

Tel: +81 3 5498 5753

E-mail: tsuji@hoshi.ac.jp

Abstract

The α3β1 integrin is an adhesion receptor for extracellular matrix proteins, and plays crucial roles in cell motility, proliferation, and differentiation. The aberrant expression of this adhesion molecule on tumor cells is frequently associated with their malignant behaviors. We previously reported that the Ets transcription factor-binding consensus sequence 133 bp upstream of the mouse α3 integrin gene is an important element for its expression in various tumor cell lines. In the present study, we attempted to identify a transcription factor bound to the Ets-consensus sequence, and found that Ets-1 bound to this sequence in an electrophoretic mobility shift assay, chromatin immunoprecipitation assay, and pull-down assay with a tandem repeat of the sequence as adsorbent. We next examined the role of Ets-1 in α3 integrin gene expression by use of a luciferase assay with a reporter plasmid containing the 5′-flanking region of the α3 integrin gene. Cotransfection of HEK293T cells with an Ets-1 expression construct and the reporter plasmid increased luciferase activity. By contrast, transfection of HT1080 cells (high α3 integrin expresser) with a dominant-negative mutant of Ets-1 decreased luciferase activity. Overexpression of Ets-1 in HepG2 hepatocellular carcinoma cells (low α3 integrin expresser) upregulated α3 integrin expression as assessed by immunoprecipitation. Finally, the induction of α3 integrin gene expression in HepG2 cells after transforming growth factor-β1 treatment was abrogated by the dominant-negative mutant of Ets-1. These results suggest that Ets-1 is involved in transcriptional activation of the α3 integrin gene through its binding to the Ets-consensus sequence at −133 bp.

Abbreviations
ChIP

chromatin immunoprecipitation

EMSA

electrophoretic mobility shift assay

Ets-1-DN

dominant-negative mutant of Ets-1

HCC

human hepatocellular carcinoma

HRP

horseradish peroxidase

MMP

matrix metalloproteinase

TGF

transforming growth factor

Introduction

Integrins are a family of cell adhesion molecules that mediate cell–extracellular matrix and cell–cell interactions, and are involved in various physiologically important processes, such as cell growth, motility, and differentiation. These molecules consist of noncovalently associated heterodimers (α and β subunits), and their combination defines the binding specificity of individual integrins for counter-ligands. The α3β1 integrin (CD49c/CD29, VLA-3, galactoprotein b3) has been identified as a high-affinity receptor for isoforms of laminin, i.e. laminin-5 (laminin-332) and laminin-10/11 (laminin-511/521) [1, 2]. This adhesion molecule was also reported to serve as a receptor for several extracellular matrix proteins, such as fibronectin, collagen, laminin-1 (laminin-111), and thrombospondin-1 [3, 4], suggesting that α3β1 integrin possesses pleiotropic binding specificities. The α3β1 integrin has a wide distribution in various epithelial and endothelial tissues, and plays important roles in the organization and maintenance of these tissues. Knockout of the α3 integrin subunit gene in mice results in abnormalities of the lung, kidney, and skin, suggesting that this integrin plays essential roles in their development and differentiation [5]. The cDNAs for the hamster, human and mouse α3 integrin subunits have been cloned [6-9].

Oncogenic transformation of fibroblastic cells by SV40 or polyoma virus increased expression of α3β1 integrin [8]. The increased expression of α3β1 integrin in the transformed cells is considered to be closely related to their oncogenic phenotypes. Many studies have demonstrated that the aberrant expression of this integrin in tumor cells of different tissue origins is associated with changes in their invasive and metastatic potentials. Expression of α3β1 integrin in various tumor cells, including gastric carcinoma, hepatocellular carcinoma, melanoma, and glioma, was positively correlated with their malignancy [10-15]. Previous studies, including ours, have indicated that expression of α3β1 integrin on gastric carcinoma cells is implicated in the formation of peritoneal metastasis [16, 17]. It was also reported that α3β1 integrin mediated activation signals for the production of matrix-degrading enzymes, such as matrix metalloproteinases (MMPs), in tumor cells [18-20]. Thus, the regulatory mechanism for α3β1 integrin expression in tumor cells seems to be of considerable interest in cancer biology.

We previously characterized the promoter region of the mouse α3 integrin subunit gene, and found that the putative binding site for Ets transcription factors located 133 bp upstream of the major transcription start site is crucial for the expression of the α3 integrin subunit gene in various tumor cell lines, including gastric carcinoma and melanoma cell lines [21]. We also reported that the putative binding site for the Sp family of transcription factors at −69 bp plays a cooperative role in the expression of the α3 subunit integrin gene in some tumor cell lines [22]. In the present study, we attempted to identify a transcription factor(s) that bound to the Ets-consensus sequence at −133 bp of the α3 integrin gene, and found that Ets-1, a prototype member of the Ets family of transcription factors, bound to the consensus sequence; we then evaluated its role in gene expression.

It has been suggested that transforming growth factor (TGF)-β is a key cytokine in regulation of the expression of α3β1 integrin in human hepatocellular carcinoma (HCC) cells, and stimulates the invasive phenotypes of HCC cells [23]. We previously reported that the Ets-consensus sequence at −133 bp is also important for expression of the α3 integrin gene induced by TGF-β1 [24]. In this study, therefore, we also examined the contribution of Ets-1 to the TGF-β1-induced transcriptional activation of the α3 integrin gene.

Results

Identification of a transcription factor bound to the Ets-consensus sequence

Our previous studies indicated that an approximately 0.3-kb fragment upstream of the α3 integrin gene (L0.3, −260/+84) showed strong promoter activity in various tumor cell lines, and that the Ets transcription factor-binding consensus sequence at −133 bp is the most important element for transcriptional control of the α3 integrin subunit (Fig. 1) [21, 22]. In this study, we first tried to isolate a transcription factor bound to the −133 bp Ets-consensus sequence from a nuclear extract of MKN1 gastric carcinoma cells. The proteins bound to Sepharose conjugated to the DNA containing a tandem repeat of the Ets-consensus sequence were analyzed by SDS/PAGE followed by silver staining. As shown in Fig. 2A, a protein of approximately 50–55 kDa was specifically bound to the wild-type sequence among several proteins bound by both wild-type and mutant DNA beads. Considering the estimated molecular mass of this protein, we analyzed the same specimens by immunoblotting with antibodies against Ets-1 and Ets-2, both of which are members of the Ets transcription factor family with molecular masses of 50–55 kDa [25]. Antibody against Ets-1 detected a protein band of 55 kDa in the eluate from the wild-type oligonucleotide, but not in that from the mutant oligonucleotide (Fig. 2B). However, no protein band was observed when antibody against Ets-2 was used in the immunoblotting assay (data not shown).

Figure 1.

Structure of the 5′-flanking region of the mouse α3 integrin gene. The map shows the location of the Ets transcription factor-binding consensus sequence at −133 bp and the nucleotide sequence surrounding the Ets-binding site. The wild-type DNA fragment (L0.3, −260/+84) shows high promoter activity, but the mutant at the Ets-consensus site (mE) loses the activity [22]. The positions for the restriction endonucleases (Sal I and Sac I) and exon I are also shown. The translation initiation site is indicated by ATG.

Figure 2.

Analysis of nuclear proteins bound to the Ets-consensus sequence. Nuclear proteins from MKN1 cells were mixed with streptavidin–Sepharose conjugated with a biotinylated tandem repeat of double-stranded oligonucleotide containing the wild-type Ets-consensus sequence (WT) or the mutated oligonucleotide at the Ets-site (Mutant). The proteins bound to the DNA-conjugated Sepharose were separated by SDS/PAGE (10%), and detected by silver staining (A) and by immunoblotting with antibody against Ets-1 (B).

We then performed an electrophoretic mobility shift assay (EMSA) with an oligonucleotide containing the Ets-consensus sequence as a probe and a nuclear extract of MKN1 cells, and found three specific DNA·protein complexes (Fig. 3A, lane 3), confirming the results of our previous study [23]. However, no specific DNA·protein complex was detected when a mutant oligonucleotide was used as a probe (Fig. 3A, lane 4) or when the wild-type competitor was added to the reaction mixture (Fig. 3A, lane 5). By contrast, the addition of a mutant competitor did not change the electrophoretic profile (Fig. 3A, lane 6). We then conducted a supershift assay with antibodies against Ets-1 and Ets-2. The mobility of one of the DNA·protein complexes was specifically retarded when antibody against Ets-1 was added to the reaction mixture before electrophoresis (Fig. 3A, lane 2), whereas the supershift was not observed with antibody against Ets-2 (Fig. 3A, lane 1). These results strongly suggest that Ets-1 specifically bound to the Ets-consensus sequence in the promoter region of the α3 integrin subunit gene. We confirmed the binding of Ets-1 to the Ets-consensus site at −133 bp of the α3 integrin gene with a chromatin immunoprecipitation (ChIP) assay with antibody against Ets-1. A larger amount of the PCR product was detected in agarose gel electrophoresis when the DNA fragments that coimmunoprecipitated with antibody against Ets-1 were used as templates for PCR than when control IgG was used (Fig. 3B).

Figure 3.

EMSA and ChIP assay. (A) A 32P-labeled oligonucleotide probe containing the Ets-consensus sequence at −133 bp was incubated with a nuclear extract from MKN1 cells for 20 min at room temperature, and the mixture was subjected to PAGE. W, wild-type oligonucleotide probe (−147/−119); m, a mutant at the Ets-consensus sequence. Three bands of protein·oligonucleotide complexes specific for the wild-type probe are indicated by closed arrowheads (lane 3). No specific band was observed when the mutant probe was used (lane 4). For competition analysis, a 50-fold molar excess of unlabeled oligonucleotides was added to the incubation mixture (lane 5, competition with the wild type; lane 6, competition with the mutant). A supershift assay was conducted with antibody against Ets-1 (lane 2) or antibody against Ets-2 (lane 1). The wild-type probe without a nuclear extract was run as a control (lane 7). (B) A ChIP assay with antibody against Ets-1 was performed. Chromatin fragments from MKN1 cells were incubated with antibody against Ets-1 or control rabbit IgG, and the immune complex was recovered with Protein G–Sepharose. The DNA fragment coprecipitating with antibody against Ets-1 was amplified by PCR as described in Experimental procedures, and the products were electrophoresed on a 2% agarose gel; this was followed by ethidium bromide staining. Chromatin fragments without immuno-precipitation were also used as a template for PCR (Total).

Effect of Ets-1 on expression of the α3 integrin subunit gene

We next conducted a luciferase assay to examine the effect of Ets-1 on expression of the α3 integrin subunit gene. Cotransfection of HEK293T cells with an Ets-1 expression construct and a reporter plasmid containing the 5′-flanking region of the α3 integrin gene (L0.3) resulted in a 2.6-fold stimulation of luciferase activity (Fig. 4A). However, introduction of an Ets-2 expression construct or a control vector did not stimulate luciferase activity. When we introduced a reporter plasmid with a mutation in the Ets-consensus sequence at −133 bp, instead of the wild-type reporter plasmid (L0.3), we observed no significant increase in luciferase activity after transfection with Ets-1 cDNA (Fig. 4A). The enhancement of luciferase activity induced by introduction of the Ets-1 expression construct depended on the dose of the Ets-1 construct used for transfection, and reached the maximum level (3.2-fold stimulation) when 400 ng of DNA was used for the transfection (Fig. 4B). We also confirmed expression of Ets-1 at the protein level in the transfectant cells by immunoblotting analysis (Fig. 4C). Similar experiments on the overexpression of Ets-1 using HepG2 cells, instead of HEK293T cells, gave essentially similar results (data not shown).

Figure 4.

Effect of expression of Ets-1 on the promoter activity of the α3 integrin gene. (A) HEK293T cells were transfected with the α3 integrin reporter plasmids (50 ng) together with the Ets-1 or Ets-2 expression construct or a control vector (pcDNA3.1 V5 HisA) (400 ng), by lipofection. After the cells had been cultured for 48 h, the luciferase activity of the cell lysate was determined. The luciferase activity was normalized to the β-galactosidase activity induced by cotransfection with the pRSV–β-Gal plasmid (50 ng). The assays were carried out in triplicate, and the data are presented as the mean ± standard deviation. L0.3, the wild type (−260/+84); mE, the construct of L0.3 mutated at the Ets-consensus site at −133 bp. (B) HEK293T cells were cotransfected with the α3 integrin reporter plasmid L0.3 (50 ng) and the Ets-1 construct (0–400 ng). The relative luciferase activity was determined as described above. The total amounts of DNA for the transfection were kept at 450 ng in each specimen by adding an appropriate amount of the control plasmid, i.e. [Ets-1 expression plasmid (x ng)] + [control plasmid (400 − x ng)] + [reporter plasmid (50 ng)] = 450 ng. (C) The expression level of Ets-1 in the HEK293T transfectant cells was analyzed by immunoblotting. The specimen was subjected to SDS/PAGE (10%) and immunoblotting with antibody against Ets-1 or antibody against β-actin. (D) Effect of Ets-1-DN on the promoter activity of the α3 integrin gene. HT1080 cells were cotransfected with the α3 integrin reporter plasmid (L0.3 or mE, 50 ng each) and Ets-1-DN or control empty vector (pcDNA3.1 V5 HisA, 400 ng), and luciferase activity was measured after incubation for 48 h as described above. (E) The dose response of Ets-1-DN to the promoter activity was examined. The procedures were essentially identical to those in (D), except that the transfection of HT1080 cells was performed with various amounts (0–400 ng) of Ets-1-DN. The total amounts of DNA for the transfection were adjusted by adding the control plasmid as described in (B).

Transfection of HT1080 cells, which express high levels of α3β1 integrin, with a dominant-negative mutant of Ets-1 (Ets-1-DN) decreased luciferase activity to 52% as compared with the control mock transfection. The decreased luciferase activity was similar to that seen in HT1080 cells transfected with a reporter plasmid with a mutation in the Ets-consensus sequence at −133 bp (Fig. 4D). The suppression of luciferase activity was found to be dependent on the amounts of Ets-1-DN used for the transfection (Fig. 4E). Thus, Ets-1 was involved in transcriptional regulation of the α3 integrin subunit gene through its binding to the Ets-consensus sequence at −133 bp in the promoter region.

We then analyzed the expression of α3β1 integrin at the protein level after overexpression of Ets-1 in HepG2 cells (low-level expressers of α3 integrin). HepG2 cells were transfected with the Ets-1 expression construct, and selected with geneticin (0.8 mg·mL−1) for 2 weeks. The transfection and selection were conducted with two separate batches. Immunoblotting analysis (Fig. 5A) and luciferase assay (Fig. 5B) confirmed that expression of Ets-1 and transcriptional activation of the α3 integrin gene in the transfectant cells of both batches (1 and 2) were increased after the cDNA transfection as compared with those in mock transfectants. The expression of α3β1 integrin on the cell surface was assessed by immunoprecipitation following surface labeling with biotin (Fig. 5C). The result indicated that Ets-1-transfected HepG2 cells had higher expression of α3β1 integrin on their cell surface.

Figure 5.

Effect of Ets-1 cDNA transfection on expression of α3 integrin. HepG2 cells were transfected with an Ets-1 expression construct, and selected with geneticin (0.8 mg·mL−1) for 2 weeks. The transfection was conducted with two batches for the full-length Ets-1 cDNA and one batch for a mock transfection. (A) The expression level of Ets-1 protein in the HepG2 transfectant cells was analyzed by immunoblotting with antibody against Ets-1. Antibody against β-actin was used as a control. (B) The stable Ets-1 transfectants of HepG2 cells were further transfected with the α3 integrin reporter plasmid (L0.3) (400 ng), and assayed for luciferase activity after being cultured for 48 h. The luciferase activity was normalized to β-galactosidase activity induced by cotransfection with the pRSV–β-Gal plasmid (100 ng). The assays were carried out in triplicate, and the data are presented as the mean ± standard deviation. (C) The cell surface proteins of HepG2 transfectant cells were labeled with biotin, and α3β1 integrin was immunoprecipitated with the procedures described in Experimental procedures. After the precipitate had been separated by SDS/PAGE (7.5%), biotin-labeled proteins blotted on the membrane were detected with HRP–streptavidin and the ECL detection system.

Involvement of Ets-1 in α3 integrin upregulation in HepG2 cells induced by TGF-β1

HepG2 cells were cultured with TGF-β1 (0.5 ng·mL−1) for 72 h, and the change in the expression level of α3 integrin was analyzed by flow cytometry. Expression of α3 integrin was increased after TGF-β1 stimulation (Fig. 6A), confirming previous observations, including ours [23, 24]. We then performed a luciferase assay to examine whether Ets-1-DN affects the promoter activity of the α3 integrin gene. Treatment of control HepG2 cells with TGF-β1 increased luciferase activity, indicating that TGF-β1 activated the promoter of the α3 integrin gene, whereas TGF-β1 treatment of HepG2 cells that had been transfected with Ets-1-DN showed almost no enhancement of luciferase activity (Fig. 6B). We then analyzed expression of α3 integrin at the protein level by flow cytometry after transfection of HepG2 cells with Ets-1-DN. The increase in expression of α3 integrin induced by TGF-β1 was partly suppressed by the introduction of Ets-1-DN (Fig. 6C). These results suggest that the α3 integrin expression induced by TGF-β1 stimulation is mediated by transcriptional activation through Ets-1.

Figure 6.

Involvement of Ets-1 in the α3 integrin gene expression induced by TGF-β1. (A) The alteration of α3 integrin expression in HepG2 cells after TGF-β1 stimulation was analyzed by flow cytometry. HepG2 cells were cultured in the presence or absence of TGF-β1 (0.5 ng·mL−1) for 72 h, and applied to a flow cytometer (FACS Calibur) after being stained with antibody against α3 integrin and fluorescein isothiocyanate-labeled secondary antibodies. The control profiles without antibody against α3 integrin are indicated by lines without shading. (B) The effect of Ets-1-DN on the promoter activity of the α3 integrin gene induced by TGF-β1 stimulation was examined. HepG2 cells were cotransfected with the α3 integrin reporter plasmid (L0.3) (50 ng) and Ets-1-DN or a control vector (pcDNA3.1 V5 HisA) (400 ng), and cultured with or without TGF-β1 (0.5 ng·mL−1) for 72 h. The cell lysate was assayed for luciferase activity. The luciferase activity was normalized to the β-galactosidase activity induced by cotransfection with the pRSV–β-Gal plasmid (50 ng). The assays were carried out in triplicate, and the data are presented as the mean ± standard deviation. (C) The effect of transfection with Ets-1-DN on the TGF-β1-induced increase in α3 integrin expression in HepG2 cells was examined by flow cytometry. HepG2 cells were transfected with Ets-1-DN or control construct, and cultured in the presence of geneticin (0.8 mg·mL−1) for 2 weeks. These cells were then treated with TGF-β1, and analyzed by flow cytometry as described above.

Discussion

We have previously reported that the Ets-consensus sequence at −133 bp in the promoter region of the α3 integrin gene is the most important element involved in regulating transcription of the α3 integrin gene [21, 22]. In the present study, we found that the transcription factor Ets-1 bound to the Ets-consensus sequence in a pull-down assay, an EMSA, and a ChIP assay (Figs 2 and 3). The overexpression of Ets-1 in cells with low expression of α3 integrin resulted in transcriptional activation of the α3 integrin gene (Fig. 4) and upregulation of α3β1 integrin on the cell surface (Fig. 5). By contrast, introduction of the Ets-1-DN into cells with high expression of α3 integrin suppressed transcription of the gene in a dose-dependent manner (Fig. 4E). These results strongly suggest that Ets-1 is involved in transcriptional regulation of the α3 integrin gene.

Ets-1 promotes the transcription of various molecules related to wound healing, angiogenesis, and tumor invasion, such as MMPs, urokinase-type plasminogen activator, and vascular endothelial cell growth factor and its receptor [25-29]. In addition, it has been reported that expression of Ets-1 is associated with tumor progression of various carcinomas and poor prognosis of patients [30, 31]. Enhanced expression of Ets-1 might promote malignant behaviors of tumor cells through upregulation of α3β1 integrin as well as through upregulation of these invasion/metastasis-related molecules. It has been reported that the enhanced expression of α3 integrin observed in highly invasive breast and gallbladder carcinoma cells is associated with upregulation of Ets-1 [32, 33].

The consensus sequence for Ets transcription factors was found in a number of integrin gene promoters, e.g. the promoters of integrin subunits α4 [34], α5 [35], αM [36], αIIb [37, 38], αV [39], and β2 [40]. Ets-1 was reported to actually regulate the expression of β3 integrin in endothelial cells [41], α10 integrin in chondrocytes [42] and αV integrin in melanoma cells [43]. The integrin family of cell adhesion molecules consists of 24 distinct α/β heterodimers that have been identified so far, and the individual members have specific cellular/tissue distributions. One of the reasons why each type of cell possesses a specific profile of integrin expression may be that the expression of each integrin subunit is regulated by several transcription factors. We previously reported that α3 integrin gene expression in certain types of cell was cooperatively regulated by the Ets-consensus and Sp-consensus sequences in the promoter region, and identified Sp3 as a transcription factor bound to the Sp-consensus sequence [22]. Kita et al. reported that the stable transfection of a glioma cell line with a dominant-negative form of Ets-1 reduced fibronectin-mediated cell adhesion and migration [44]. Although it was expected that expression of α5β1 integrin, a high-affinity receptor for fibronectin, would be affected by the dominant-negative transfection, we did not observe upregulation of this integrin in HepG2 cells after Ets-1 overexpression (unpublished observation). It is likely that the transfection causes the alteration in the expression profile of various adhesion molecules, including α3β1 integrin.

TGF-β1 is a multifunctional cytokine that regulates various physiologically important molecules, including cell adhesion molecules [45]. Giannelli et al. noted that TGF-β1 stimulated noninvasive hepatocellular carcinoma cells to acquire invasive potential that accompanied the upregulation of α3 integrin [23]. Our research group then suggested that the Ets-binding consensus sequence at −133 bp is an element that responds to TGF-β1 stimulation [24], and that Ets-1 was involved in the augmentation of α3 integrin expression induced by TGF-β1 in the present study (Fig. 6). The major pathway of intracellular signaling transduced by TGF-β1 is known to be mediated by the so-called Smad protein cascade. It was also reported that Smad proteins and Ets-1 play a cooperative role in TGF-β1 signaling for the gene expression of tenascin-C in dermal fibroblasts [46] and parathyroid hormone-related protein in breast cancer cells [47]. However, no consensus binding sequence for Smad proteins was found in the 5′-flanking region of the α3 integrin gene [21], suggesting that Smad-independent signaling pathways are involved in the transcriptional activation of α3 integrin. One possibility is the involvement of mitogen-activated protein kinase-dependent activation of members of the Ets family of transcription factors. Recent studies have suggested that the mitogen-activated protein kinase cascades mediate the TGF-β1-stimutated upregulation of MMP-9 in glomerular podocytes [48]. Another possibility may be the induction of the transcription factor Snail, which is upregulated by TGF-β1 treatment of highly invasive squamous cell carcinoma [49]. The enhanced expression of Snail resulted in an increase in Ets-1 expression and its DNA-binding activity in these cells, causing MMP-2 expression through promoter activation. The intracellular events linking TGF-β stimulation and Ets-1-driven α3 integrin gene expression remain to be elucidated in a future study. In conclusion, the present study strongly suggests that Ets-1 is implicated in the expression of α3 integrin in tumor cells via its binding to the 133 bp region upstream of the α3 integrin subunit gene. Possible cooperation between Ets-1 and other transcription factors, such as Sp3, is also an important issue to be clarified.

Experimental procedures

Cells

MKN1 cells (human gastric carcinoma cell line) were provided by the RIKEN cell bank (Tsukuba, Japan). HepG2 cells (human hepatocellular carcinoma cell line) and HT1080 cells (human fibrosarcoma cell line) were provided by the Human Science Research Resources Bank (Osaka, Japan). HEK293T cells (human embryonic kidney cell line) were donated by A. Muto (Hoshi University). These cells were cultured in RPMI-1640 medium (Gibco BRL, Rockville, MD, USA) supplemented with 10% fetal bovine serum at 37 °C under a 5% CO2 atmosphere.

Reagents and antibodies

Restriction endonucleases and modifying enzymes were purchased from TaKaRa (Osaka, Japan), TOYOBO (Osaka Japan), and Gibco BRL. Ex Taq polymerase was obtained from TaKaRa. Trizol and Super Script II Reverse Transcriptase were purchased from Invitrogen (Carlsbad, CA, USA). Geneticin (G418) was obtained from Gibco BRL. p-Nitrophenyl-β-d-galactopyranoside was obtained from Sigma (St Louis, MO, USA). Oligonucleotides were supplied by Hokkaido System Science (Sapporo, Japan). [32P]ATP[γP] was supplied by Perkin-Elmer (Waltham, MA, USA). Streptavidin–Sepharose and Protein G–Sepharose were purchased from GE Healthcare (Piscataway, NJ, USA). Biotin-NHS-WS was obtained from Vector Laboratories (Burlingame, CA, USA). Recombinant human TGF-β1 was purchased from PeproTech (Rocky Hill, NJ, USA). A transfection reagent, cationic hydroxyethylated cholesterol-based nanoparticle, was prepared as described previously [50].

Monoclonal antibody against α3 integrin (clone SM-T1) was prepared in our laboratory [51]. Antibodies against Ets-1 and Ets-2 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Antibody against β-actin was obtained from Sigma. Fluorescein isothiocyanate-labeled anti-mouse IgG and horseradish peroxidase (HRP)-conjugated anti-mouse and rabbit IgG were purchased from Kirkegaard & Perry Laboratories (Gaithersburg, MD, USA). HRP-conjugated streptavidin was purchased from Vector Laboratories.

Isolation of sequence-specific DNA-binding proteins

The proteins bound to the Ets-consensus sequence of the α3 integrin gene promoter were isolated by affinity purification. The affinity resin with an oligonucleotide containing a tandem repeat of the Ets-consensus sequence was synthesized according to the method of Kadonaga and Tjian [52], with a slight modification [22]. A pair of complementary oligonucleotides with the following sequences were designed to have cohesive ends after they are annealed: 5′-TTTTCTCTTTCCCCGGAAGGAAAGCAGAGCGGTG-3′ (G1) and 5′-CTCTGCTTTCCTTCCGGGGAAAGAGA AAACACCG-3′ (G2) (complementary sequences are underlined) (Fig. 1). G1 and G2 oligonucleotides were annealed and ligated to form an oligomer, and the double-stranded oligonucleotide products were subcloned into a T-easy vector (Promega, Madison, WI, USA). For the control experiment, a mutated double-stranded oligonucleotide was also prepared with the following complementary oligonucleotides, using the same procedure: 5′-TTTTCTCTTTCCCCGTAAGGAAAGCAGAGCGGTG-3′ (mG1) and 5′-CTCTGCTTTCCTTACGGGGAAAGAGAAAACACCG-3′ (mG2) (mutated bases are underlined). From the wild-type and mutant libraries, we picked up one clone each with an approximately 1.0-kb insert that corresponded to ~ 30 repeats of the G1/G2 or mG1/mG2 units. Subsequently, a biotinylated double-stranded oligonucleotide was prepared by PCR, with a plasmid containing the tandem repeat as a template, and a set of primers with the following sequences: 5′-(biotin)-ATGGCGGCCGCGGGAATT-3′ and 5′-AGGCGGCCGCGAATTCACTA-3′. The conditions for PCR were: 95 °C for 1 min; 55 °C for 1 min; 72 °C for 3 min; 10 cycles. The PCR product was conjugated with streptavidin–Sepharose. The DNA-conjugated Sepharose thus prepared was mixed with a nuclear extract (0.8 mg of protein, 0.5 mL) prepared from MKN1 cells as described by Ko et al. [53] in binding buffer (25 mm Tris/HCl, pH 7.9, 65 mm KCl, 6 mm MgCl2, 0.25 mm EDTA, and 10% glycerol) containing poly(dI-dC) (20 μg), and the mixture was incubated for 24 h at 4 °C. After DNA-conjugated Sepharose had been washed with binding buffer and then with 0.25 m KCl, the bound proteins were eluted by treatment with sample buffer for SDS/PAGE (0.2 m Tris/HCl, pH 6.8, 2% SDS, 10% glycerol, and 0.02% bromophenol blue) at 95 °C for 3 min. The DNA-binding proteins were separated by SDS/PAGE (10%). The gel was stained with a silver staining kit (2D Silver Stain-II; Cosmo Bio Co., Tokyo, Japan).

Immunoblotting

The proteins separated by SDS/PAGE were blotted onto a nitrocellulose membrane with a semi-dry blotting system as described previously [54]. The membrane was treated successively with antibody against Ets-1 (1 : 1000 dilution) or antibody against Ets-2 (1 : 1000 dilution) for 20 h, and with HRP-conjugated anti-mouse IgG (1 : 1000 dilution) for 40 min. Proteins were detected with the ECL detection system (GE Healthcare).

Ets-1 in the transfectant cells was also analyzed by immunoblotting, essentially under the same conditions as those described above. The sample was prepared as follows: transfectant cells were incubated in RIPA buffer (25 mm Tris/HCl, pH 7.6, 150 mm NaCl, 1% Nonidet P-40, 0.1% SDS) at 4 °C for 15 min, and the lysate was centrifuged at 15 000 g for 15 min. The supernatant was mixed with an equal volume of sample buffer for SDS/PAGE.

EMSA

The double-stranded oligonucleotides containing the Ets-consensus sequence (wild type and a mutant) were used as probes and competitors. The oligonucleotides used in this study were as follows: 5′-TTTTCTCTTTCCCCGGAA GGAAAGCAGAG-3′ (W1, wild-type) and 5′-TTTTCTCT TTCCCCGTAAGGAAAGCAGAG-3′ (m1, a mutant of W1) [21, 22]. The 32P-labeled probes (15 000 d.p.m.) and nuclear extracts from MKN1 cells (20 μg of protein) were mixed with the binding buffer (0.02 mL) in the presence of poly(dI-dC) (1 μg) and incubated for 20 min at room temperature. For supershift assays, 8 μg of antibody against Ets-1 or Ets-2 was added to the mixture and incubated at 4 °C for 30 min prior to the addition of the probe. The mixture was separated by PAGE (5%) with 10 mm Tris/acetate (pH 7.8) containing 0.25 mm EDTA (0.25 × Tris/acetate/EDTA) as running buffer.

ChIP

The chromatin solution was prepared from MKN1 cells (106 cells) according to the method described by Sato et al. [55]. An aliquot (1 mL) of the chromatin solution was precleared with Protein G–Sepharose/salmon sperm DNA (50 μL, 50% suspension) for 4 h at 4 °C, and incubated with antibody against Ets-1 (5 μg) at 4 °C for 16 h. The immune complex was collected by adding Protein G–Sepharose/salmon sperm DNA (30 μL, 50% suspension) and incubating at 4 °C for 2 h. The Sepharose beads were pelleted by centrifugation at 5 000 g for 10 sec, and washed sequentially with low-salt buffer, high-salt buffer, LiCl wash buffer, and TE buffer, as described previously [55]. The immune complex was eluted by treating the beads with ChIP elution buffer (10 mm Tris/HCl, pH 8.0, 300 mm NaCl, 5 mm EDTA, and 0.5% SDS, 200 μL) at 65 °C for 6 h. After the remaining RNA and proteins had been digested with RNase A (10 μg) at 37 °C for 30 min and then with proteinase K (60 μg) at 55 °C for 1 h, DNA was recovered by phenol/chloroform extraction and ethanol precipitation with glycogen (40 μg) as a carrier. The precipitated DNA was dissolved in TE buffer (20 μL) and subjected to PCR (35–39 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 7 s) with a set of primers (5′-CCGGGATCTTTTTCTCTTTC-3′ and 5′-ACCGCAAGGAGGACAGGAG-3′) to amplify the region containing the Ets-consensus sequence at −133 bp. PCR products were separated by electrophoresis on a 2% agarose gel.

Construction of plasmids for expression of Ets-1 and Ets-2

The cDNAs for human Ets-1 and Ets-2 mRNA were amplified with total RNA from MKN1 cells as a template. The following primer pairs were designed on the basis of the sequences of the human Ets-1 gene (GenBank accession no.: NM_005238) and human Ets-2 gene (GenBank accession no.: NM_005239) to isolate the full-length cDNA: Ets-1-(FL)-F (5′-TAAGCTTACTTGCTACCATCCCGTACG-3′; the HindIII site is underlined) and Ets-1-(FL)-R (5′-CGGTACCTGCCATCACTCGTCGGCATA-3′; the KpnI site is underlined); and Ets-2-(FL)-F (5′-GGGTACCGCAGCGGCAGGATGAATGAT-3′; the KpnI site is underlined) and Ets-2-(FL)-R (5′-CTGAATTCTCAGTCCTCCGTGTCGGGC-3′; the EcoRI site is underlined). For preparation of Ets-1-DN, which contains a DNA-binding domain (amino acids 306–441) but lacks a transcriptional activation domain [55], the following primers were used: Ets-1-(DN)-F (5′-GGGGTACCATGGACTATGTGCGGGACCGT-3′; the KpnI site is underlined) and Ets-1-(DN)-R (5′-GCGGCCGCGCCATCACTCGTCGGCATC-3′; the NotI site is underlined). Ets-1-(DN)-F contains an initiation codon for translation of Ets-1-DN. PCR was performed under the following conditions: 95 °C for 1 min; 55 °C for 1 min; 72 °C for 3 min; 30 cycles. PCR products were digested with appropriate restriction endonucleases, and cloned into pcDNA3.1 V5 HisA vector (Invitrogen) and sequenced.

The reporter plasmid containing the Ets-consensus sequence at −133 bp in the pGL3-basic vector (Promega) (L0.3, −260 to +84 bp of the α3 integrin subunit gene) and its mutant were prepared as described previously [21, 22]. The following primers were used to prepare the mutant at the −133-bp Ets-consensus sequence by PCR-based site-directed mutagenesis: 5′-TTTTCTCTTTCCCCGTAAGGAAAGCA-3′ and 5′-TGCTTTCCTTACGGGGAAAGAGAAAA-3′ (mutation sites are underlined). The fragments were inserted into the KpnI and SacI sites of the pGL3-basic vector.

Luciferase assay

The luciferase assay was conducted with a luciferase assay system (Promega), essentially as described previously [21]. Briefly, cells were seeded in 24-well plates (1–3 × 105 cells per well) and cultured for 24 h. The cells were transfected with a reporter plasmid (50 ng) and an expression construct of Ets-1 or Ets-2 (400 ng) by lipofection with cationic hydroxyethylated cholesterol-based nanoparticles. The cells were cotransfected with pRSV–β-Gal plasmid (50 ng) as an internal control. After the cells had been cultured for 48 h, the cell extracts were assayed for luciferase activity with a chemiluminometer. β-Galactosidase activity in the cell extract was determined to normalize the transfection efficiency, as described previously [22].

Flow cytometric analysis

Expression of the α3 integrin subunit was measured with a flow cytometer (FACS Calibur; BD Biosciences, San Diego, CA, USA), as described previously [51].

Immunoprecipitation

Cell surface proteins of HepG2 transfectant cells (107 cells) were labeled by treating the cells with biotin-NHS-WS (0.2 mg·mL−1) in NaCl/Pi at 4 °C for 20 min. The cell lysate was prepared with lysis buffer (0.05 m Tris/HCl, pH 8.0, 1% Nonidet P-40, 0.15 m NaCl, 5 mm EDTA), and this was followed by centrifugation at 15 000 g for 15 min. Antibody against α3 integrin (SM-T1) (3 μg) and Protein G–Sepharose (30 μL, 70% suspension) were added to the supernatant, and the mixture was incubated at 4 °C for 24 h with continuous gentle agitation. After Protein G–Sepharose beads had been washed three times with the lysis buffer, the bound proteins were recovered by treating the beads with the sample buffer for SDS/PAGE, and analyzed by immunoblotting as described above. The biotinylated proteins on the membrane were visualized with HRP–streptavidin (1 : 3000 dilution) treatment followed by ECL detection.

Conflict of interest

The authors have no potential conflicts of interest to disclose.

Acknowledgements

We are grateful to Y. Nakamura, M. Takahashi, A. Katsura, R. Miura and Y. Takashima (Hoshi University School of Pharmacy and Pharmaceutical Sciences) for their technical assistance. This work was supported in part by the Ministry of Education, Culture, Sports, Science and Technology of Japan, and by the Open Research Center Project.

Ancillary