Plasmid constructs.

The Wnt3A expression plasmids were a generous gift from Dr. R. Nusse. The Lef-1 promoter-luciferase reporter, LF−2700/−200-Luc, was previously described (1, 43) and encompassed −2,700 to −200 bp of the promoter relative to the translational start site of the Lef-1 protein (17). The S37A-β-catenin and mouse Sox17 (wild-type; TAD, β-catenin-binding domain deleted; t-Sox17, HMG domain deleted; Sox17mut1, HMG domain mutant M76A; and Sox17mut2, HMG domain mutant G103R) expression plasmids were previously described (41). The plasmid pUseTCF4 and the reporter TCF-responsive TOP-flash were purchased from Upstate Biotech. The TOP-flash reporter consists of tandem TCF/LEF-1 binding sites upstream of a minimal cfos promoter. The Lef-1-expressing plasmid was previously described by our laboratory (14).

Cell lines and transient transfection experiments.

The HEK293T, MCF-7, A549, SW480, and L cell cell lines were all purchased from American Type Culture Collection and grown in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. Wnt3A conditioned medium and control medium were harvested from Wnt3A-expressing L cells and control L cells, using medium with and without FBS (50). Cells were transfected using Lipofectamine-LTX (Invitrogen Life Technologies) according to the manufacturer's instructions or by electroporation for larger-scale analyses. Cells were grown to ~50–70% confluency before transfection. Unless otherwise specified, transient transfection reporter assays were performed in 24-well plates and utilized 0.2 μg of the Lef-1 promoter-luciferase reporter and 0.01 μg of the pCMV-Renilla luciferase plasmid (Promega, Madison, WI) as an internal control for transfection efficiency. For the TOP-flash reporter experiments, 24-well plates were utilized, and cells were transfected with 0.05 μg of the TOP-flash reporter and 0.01 μg of the pCMV-Renilla luciferase plasmid (internal control for normalization of transfection efficiency). In both Lef-1 promoter-luciferase reporter and TOP-flash reporter experiments evaluating the effects of other expressed proteins (0.5 μg S37A-β-catenin and/or 0.2 μg Sox17/well of a 24-well plate), a third plasmid (pcDNA3.1; Invitrogen, Carlsbad, CA) was added to the cotransfection cocktail to normalize the total amount of DNA transfected in all experimental comparisons. The transfections were carried out in serum-containing medium, and cells were harvested for the firefly and renilla luciferase assays 24–30 h after transfection unless otherwise specified. Protein was prepared for the assays by washing the cells in PBS, followed by lysis in 1× Passive Reporter Lysis Buffer (Promega). Protein concentrations were determined using the Bradford method, and all lysates were normalized to the same protein concentration using the lysis buffer. Ten micrograms of total cellular lysate was used for measurement of the relative luciferase activity units (RLU), for both firefly and renilla luciferase, using the Dual Luciferase Assay kit (Promega). Transfection efficiencies were normalized by dividing the relative firefly luciferase units by the relative renilla luciferase units. Following normalization, values were represented as RLU.

Immunoprecipitation and Western blot analyses.

Nuclear extracts and nuclear protein complexes were prepared for Western blotting and immunoprecipitation studies using the NucBuster Protein Extraction kit (Novagen, EMD Bioscience) or the Nuclear complex Co-IP kit (Active Motif, Carlsbad, CA), respectively, according to the manufacturers' recommendations. For immunoprecipitation analyses, 1,000 μg of protein was mixed with the appropriate antibodies, followed by binding to protein A or protein G agarose beads, or Nickel Dynabeads in the case of His-tagged proteins. The bound proteins were then eluted in an SDS sample buffer and resolved on SDS-polyacrylamide gels. The gels were then transferred onto Nylon C Hybond membranes, and these were blotted with appropriate antibodies. For Western blot analyses, 100 μg of nuclear complex protein was resolved by SDS-PAGE and blotted with appropriate antibodies following protein transfer to membranes. The following antibodies and sources were used for immunoprecipitation and Western blotting: antibodies against total β-catenin, active β-catenin (antibody that recognizes only β-catenin dephosphorylated on Ser37 or Thr41), Lef-1 (clone 2D12), TCF4, and GSK3-βY214 were purchased from Upstate Biotechnology. Other antibodies used in this study were anti-V5 (Invitrogen), rabbit anti-cyclin D1 (Abcam, Cambridge, MA), and goat anti-Sox17 (R&D Systems). Blots were then washed and subsequently incubated with secondary antibody (anti-IgG conjugated to HRP). The immune complexes were detected by ECL chemiluminescence or with SuperSignal (Pierce Chemical, Rockford, IL).

Induction and cell fractionation of undifferentiated primary airway epithelial cells.

Recombinant adenoviruses expressing Wnt1 or Wnt3A were generated using methods previously described (2). These recombinant adenoviral vectors (Ad.Wnt1 and Ad.Wnt3A) were used to synthesize these Wnts in HCT-116 cells. The cells were infected with 100 particles/cell, conditioned medium was collected, and Wnts were concentrated by absorption onto Blue Sepharose beads. It was necessary to concentrate the Wnts in this fashion to remove serum, which would otherwise cause primary airway epithelial cells to differentiate. Ad.BglII (empty vector control)-infected cells were used to create mock-conditioned medium. The adenovirus-containing medium was removed 16 h after infection, and the cells were washed twice with PBS before being fed a small volume of fresh medium. Conditioned medium was collected after 24 h and filtered (0.22 μm) before absorption onto Blue Sepharose beads (Amersham). Blue Sepharose beads (0.5 ml) were washed in PBS and added to 50 ml of conditioned medium. After overnight rotation of the samples at 4°C, the beads were washed in 3× PBS/1% CHAPS and stored in this solution at 4°C until use. Primary human and ferret airway epithelial cells were isolated as previously described by our laboratory and grown on plastic in a proliferative undifferentiated state (16, 33). Cells were induced by adding mock-, Wnt1-, or Wnt3A-conditioned beads to cells in culture. Thirty-six hours after the bead addition, the cells were harvested and nuclear extracts (NE) and postnuclear supernatants (PNS) were prepared using a Nuclear Extract Kit (Active Motif). Briefly, cells in a 100-mm dish were removed by gentle scraping, followed by lysis with a proprietary hypotonic lysis buffer (Active Motif). The cytoplasmic fraction (PNS) was collected after centrifuging the suspension at 14,000 g at 4°C for 30 s. The nuclear pellet was resuspended in 50 μl of proprietary complete lysis buffer (Active Motif) and incubated on ice for 30 min on a rocking platform. The sample was then vortexed for 30 s and centrifuged at 14,000 g for 10 min, after which the supernatant (NE) was transferred to a prechilled tube. Although contamination of cellular plasma membrane proteins cannot be excluded from this type of preparation, the company's lysis buffers are designed to enrich for nuclear proteins. We have previously used this method with small samples and have succeeded in enriching for nuclear and cytoplasmic proteins at >95% purity, as determined by Western blotting for H1 (nuclear) and IκBα (cytoplasmic) (52). PNS and NE were quantified for protein concentration before analysis by Western blotting for Lef-1, β-catenin, TCF-4, and Sox17.

Chromatin immunoprecipitation assays.

Three 150-mm plates of A549 cells were transfected with 30 μg of plasmid expressing either mouse Sox17 or TCF4 (41). Cells were then fixed and harvested for chromatin immunoprecipitation (ChIP) at 36 h posttransfection. Formaldehyde cross-linked DNA-protein complexes were sheared by sonication, to an average size of 400–600 bp, diluted 1:10 with the ChIP dilution buffer included in the EZChIP kit (Upstate Biologicals), and then precleared with protein A agarose beads. The precleared DNA-protein mixture was used for the ChIP assays. Ten micrograms of anti-Sox17 (R&D Systems), anti-TCF4 (Upstate Biologicals), or nonimmune control IgG antibody was added to the ChIP reaction mixture. Immunoprecipitation and DNA recovery were performed according to the EZChIP kit protocol from Upstate Biologicals. The recovered DNA was detected by PCR and quantified by real-time PCR using 3 μl of the final DNA precipitate. Ten primer sets were used to survey the 2.5-kb human Lef-1 promoter locus (see Table 3). The PCR protocol was carried out for 30 cycles, each at 95°C for 15 s and 60°C for 45 s. Real-time quantitative PCR was performed using Bio-Rad IQ SYBR Green Supermix and iCycler iQ Detection System (Bio-Rad, Hercules, CA). The relative copy number for each ChIP DNA fragment was calculated as the fold increase of the relative copy number in the immunoprecipitated fraction (with the specific capture antibody) over the relative copy number in the IgG-negative control for each sample.

Purification of His-tagged TCF4, Lef-1, and Sox17 mouse proteins.

Mouse cDNAs for TCF4, Lef-1, and Sox17 wild-type and mutant proteins (41) were cloned in-frame into the pET151/D-TOPO vector backbone using the Champion pET Directional TOPO Expression Kit (Invitrogen). The resultant clones were confirmed by DNA sequencing and expressed NH₂-terminal His-tagged fusion proteins in bacteria. The induction of correctly sized fusion proteins in bacteria was confirmed by Western blotting, using both anti-His antibody and antibodies against the protein backbone. The His-tagged fusion proteins were purified by HPLC using a HisTrap HP column (GE Healthcare Life Sciences), and the purity was confirmed by SDS-PAGE Coomassie staining and Western blotting. Concentrations of the purified proteins were normalized using the Bradford method, and aliquots were stored at −80°C until use. Purified GST and GST-β-catenin protein were purchased from Millipore (Billerica, MA).

Non-radioactive EMSA.

To test whether Sox17, TCF4, or Lef-1 protein physically binds to putative TCF/Lef-1 and/or Sox DNA binding sites in the Lef-1 promoter, we developed a non-radioactive EMSA. This approach assessed DNA/protein interactions using purified recombinant proteins and purified 5′-biotinylated double-stranded oligonucleotides (see Table 2). Five micrograms of purified His-fusion protein(s) was preincubated for 10 min at 4°C in binding buffer containing 0.005 units of poly(dG-dC) (Sigma), 0.005 units of poly(dI-dC) (Sigma), 0.05% Igepal-CA630 (Sigma), and 2 mg/ml BSA (19). After the preincubation, 2 pmol of purified double-stranded biotinylated oligonucleotide was added to the binding mixture (final volume of 20 μl) and incubated at room temperature for an additional 20–30 min. Annealed oligonucleotides were purified using a 7.5% polyacrylamide gel in 0.5× TBE buffer, and the concentration after purification was determined by slot blotting. The EMSA binding reaction mixture was resolved by electrophoresis in a 7.5% polyacrylamide gel in 0.5× TBE buffer at low voltage. After electrophoresis, the DNA was transferred to a Nylon membrane using TBE buffer at 40 V for 1–2 h. DNA was then cross-linked to the nylon membrane by UV (120 mJ/cm²) using the “autocrosslink” function on a Stratalinker. The biotinylated oligonucleotides were detected on the membrane using a NEB Phototope-Star Detection kit and following the manufacturer's instructions (New England Biolabs).

Detection of protein complexes bound to biotinylated double-stranded oligonucleotides.

To evaluate whether Sox17, TCF4, Lef-1, and β-catenin can form multiprotein complexes on DNA sequences from the Lef-1 promoter, we mixed 30 μg of purified His-tagged Sox17, TCF4, and LEF-1 proteins (~500 pmol of each) and 1.5 μg of GST-β-catenin (13 pmol) (Upstate Biologicals) in binding buffer containing 0.0025 units poly(dI-dC), 0.0025 units poly(dG-dC), 0.01 μg single-stranded (denatured) salmon sperm DNA, and 0.05% Igepal-CA630. After preincubation on ice for 10 min, 5 pmol of gel-purified biotinylated double-stranded oligonucleotides was added, and the total volume was adjusted to 200 μl with binding buffer. The molecular ratio of oligonucleotide:protein was ~1:100 for transcription factors and 1:2.5 for β-catenin. An excess of transcription factors was used to drive complex formation and allow for competitive DNA binding between transcription factors, which could only be achieved under conditions where binding was saturated. The mixture was then incubated on ice for 20–30 min, followed by the addition of 800 μl of binding buffer containing 20 μl of prewashed Streptavidin C1 DynaBeads (Invitrogen). The mixture was then gently agitated at room temperature for 30–60 min, followed by washing of the beads in a magnet with washing buffer (25 mM Tris-Cl, 150 mM NaCl, 0.1% BSA, 0.05% Tween 20, pH 7.2) five times for 3 min each. The beads were then resuspended in 100 μl 2× SDS-PAGE loading buffer and boiled for 5 min. Twenty-five microliters of the suspension was then loaded onto an SDS-PAGE gel and Western blotted with antibodies against Sox17, TCF4, Lef-1, or β-catenin.

Lef-1 expression in undifferentiated primary airway epithelial cells is induced by Wnt3A and Wnt1, and this induction correlates with a decline in expression of full-length Sox17.

Both Wnt1 and Wnt3A have the ability to induce the Lef-1 promoter in transformed cell lines (11, 17) and glandular progenitor cells (10). We evaluated whether beads conjugated to Wnt1 or Wnt3A proteins (Fig. 2A) could modulate Lef-1 and/or Sox17 expression in undifferentiated primary human and ferret tracheobronchial airway epithelial cells (AECs). Indeed, treatment with either Wnt1- or Wnt3A-conjugated beads resulted in higher levels of Lef-1 and TCF4 protein expression than did treatment with unconjugated (mock) beads (Fig. 2B). Similar findings of Wnt3A-mediated induction of Lef-1 and TCF4 were seen in both human and ferret undifferentiated primary AECs treated with conditioned medium from Wnt3A-producing L cells (Fig. 2C). In these studies, Lef-1 induction correlated with a rise in cyclin D1 expression, consistent with previous findings that Lef-1 binds to and induces the cyclin D1 promoter (40) and that Lef-1 knockout mice have defects in glandular progenitor cell proliferation including cyclin D1 expression (10). Subcellular fractionation experiments demonstrated that Wnt1 and Wnt3A induced Lef-1, but not β-catenin, in the nuclear fraction and that the ratios of full-length (F-Sox17) and HMG-deleted (t-Sox17, which does not bind DNA) isoforms of Sox17 changed in response to Wnt induction. F-Sox17 levels decreased in the presence of Wnt while t-Sox17 levels increased (Fig. 2D). Because the F-Sox17 isoform was only weakly expressed in this model system relative to t-Sox17, the migratory position of the F-Sox17 protein within the SDS-PAGE was confirmed by recombinant adenovirus overexpression of F-Sox17 in undifferentiated primary AECs (Fig. 2E). Cumulatively, these results demonstrate that Wnts can induce Lef-1 and cyclin D1 expression in primary airway epithelial cells while simultaneously reducing the expression of F-Sox17. This phenotype is similar to that observed in glandular progenitor cells at the time of bud formation (10) (Fig. 1).

Open in a separate window

Fig. 2.

Wnt proteins stimulate Lef-1 expression in undifferentiated primary airway epithelial cells. A: HCT-116 cells were infected with the indicated recombinant adenoviral vectors expressing Wnt1 or Wnt3A (Ad.Wnt1 and Ad.Wnt3A) or with a control adenoviral vector (Ad.BglII) lacking a transgene. Conditioned HCT-116 medium was absorbed to Blue Sepharose beads, and 20 μl of the 50-ml conditioned medium (m) or 20 μl of the 500 μl of absorbed beads (b) were loaded in the indicated lanes and subjected to Western blotting with anti-Wnt3A or anti-Wnt1 antibodies as indicated. B: undifferentiated primary human bronchial airway epithelial cells (AECs) grown on plastic were treated with the indicated Wnts absorbed to beads. At 36 h following treatment, whole cell lysates were harvested for Western blotting with anti-Lef-1, anti-TCF-4, and anti-β-catenin antibodies (noted at right). C: conditioned medium harvested from Wnt3A-expressing transgenic L cells or from control L cells lacking the Wnt3A transgene was applied to undifferentiated primary human (left) or ferret (right) AECs for 72 h. Nuclear extracts were used for Western blotting with anti-Lef-1, anti-β-catenin, anti-TCF4, anti-cyclin D1, and anti-β-actin antibodies. D: undifferentiated primary AECs were treated with Wnt-conjugated beads. At 36 h following treatment, postnuclear supernatants (PNS) and nuclear extracts (NE) were generated and evaluated by Western blotting with anti-Lef-1, anti-Sox17, and anti-β-catenin antibodies. F-Sox17 (long isoform) and t-Sox17 (truncated short-form missing the HMG binding domain) are shown, at long and short exposure times, respectively. E: a recombinant adenoviral vector encoding the long form of Sox17 (F-Sox17) was used to demonstrate isoform migration. Lane 1, nuclear extract from mock-treated undifferentiated primary AECs (from D) for comparison of F-Sox17 and t-Sox17 expression (long exposure time); lanes 2 and 3, Ad.BglII-infected AECs; lanes 4 and 5, Ad.F-Sox17-infected AECs.

Sox17 modulates Wnt3A-mediated activation of the Lef-1 promoter.

We next tested whether Sox17 expression can modulate Wnt-mediated activation of the Lef-1 promoter, using transient transfection of 293 cells with a 2.5-kb Lef-1 promoter-driven luciferase reporter. As a control for Wnt/TCF-mediated activation, we also evaluated the well-characterized Wnt/TCF-responsive reporter TOP-flash, whose activation by β-catenin is inhibited by Sox17 (41, 54). Results from this analysis demonstrated that the Lef-1 promoter was dynamically induced by Wnt3A expression, with activity peaking at 24–48 h posttransfection and declining thereafter (Fig. 3A). Coexpression of Sox17 with Wnt3A significantly reduced Wnt3A-mediated activation of the Lef-1 promoter. Paradoxically, however, Sox17 expression alone resulted in levels of Lef-1 promoter activity above baseline (Fig. 3, A and C). In the case of the TOP-flash reporter, Sox17 expression inhibited Wnt3A-mediated activation, as previously reported for its activation by Wnt1 or dominant-activated β-catenin (41, 54); notably, the expression of Sox17 alone did not induce TOP-flash expression as had been observed for the Lef-1 promoter (Fig. 3D). Immunoblotting analysis revealed that Sox17 expression promoted the degradation of activated β-catenin (dephosphorylated on Ser37 or Thr41) in both the presence and absence of Wnt3A, without altering the level of tyrosine-phosphorylated PY-214-GSK-3β (Fig. 3B). These observations are consistent with previous findings that Sox17 can inhibit Wnt signaling by stimulating GSK3β-independent degradation of β-catenin protein in colon cancer cell lines (12, 41).

Open in a separate window

Fig. 3.

Sox17 dynamically regulates the Lef-1 promoter in the presence and absence of Wnt3A. A: a 2.5-kb Lef-1 promoter-luciferase reporter (Lef-1-Luciferase) construct was used to evaluate the kinetics of promoter activation in the presence of Wnt3A and/or Sox17 expression. 293 Cells were cotransfected with 0.2 μg of Lef-1-luciferase reporter and 0.5 μg of Wnt3A and/or 0.2 μg of Sox17 expression plasmids using Lipofectamine LTX reagent. For all transfections, the total DNA concentration was identical and normalized using the empty-vector control plasmid pcDNA. All transfections also contained 0.01 μg of a pCMV-Renilla luciferase plasmid for normalization of transfection efficiency. Relative activity of the Lef-1 promoter is plotted as relative light units (RLU). B: Western blot analysis of changes in total β-catenin, active β-catenin, Sox17, and phosphorylated GSK3β at 24 h posttransfection, with the transfected plasmids indicated above the blot. C and D: comparison of Lef-1-luciferase reporter (0.2 μg) (C) and TOP-flash TCF-responsive reporter (0.05 μg) (D) activities in cells transfected with the indicated reporter and Wnt3A (0.5 μg) and/or Sox17 (0.2 μg) expression constructs. E and F: comparison of Lef-1-luciferase reporter (E) and TOP-flash TCF-responsive reporter (F) activities in cell transfected with the indicated reporters and dominant-active β-catenin-S37A (0.5 μg) and/or Sox17 (0.2 μg) expression constructs. Values in A and C–F represent the mean (± SE, n = 9) RLU normalized for Renilla luciferase expression.

To further investigate both the importance of activated β-catenin in regulating the Lef-1 promoter and the ability of Sox17 to inhibit this activation, we studied induction of the Lef-1 promoter in 293 cells expressing dominant-active S37A-β-catenin and/or Sox17. Like Wnt3A, S37A-β-catenin activated both the Lef-1 promoter and TOP-flash reporters (Fig. 3, E and F). However, the extent of Sox17-mediated repression was significantly less than that seen in cells transfected with Wnt3A (Fig. 3, C and D). These findings suggested that Sox17-mediated repression of the Wnt3A-activated Lef-1 and TOP-flash promoters likely involves additional Wnt3A-induced factors besides β-catenin.

Sox17 regulates Wnt3A-mediated activation of Lef-1 promoter in a cell type-dependent fashion.

To substantiate results in 293 cells, we performed promoter-reporter experiments in four additional cell lines: MCF7 (human mammary adenocarcinoma), SW480 (human colon adenocarcinoma), A549 (human lung carcinoma), and L cells (murine transformed fibroblast). In the MCF-7, SW480, and A549 cells, the ability of Sox17 to repress Wnt3A-mediated activation of the Lef-1 and TOP-flash promoters was similar to that seen in 293 cells (Fig. 4). In L cells, by contrast, Wnt3A failed to induce the Lef-1 promoter and only weakly induced the TOP-flash reporter; however, in the absence of a Wnt signal, Sox17 expression led to repressed baseline activities of both these promoters. Notably, none of the four cell lines demonstrated Sox17-mediated activation of the Lef-1 promoter in the absence of Wnt3A, as had been observed in 293 cells. These data suggest that the ability of Sox17 to inhibit and/or activate the Lef-1 promoter in the presence or absence of a Wnt signal is likely highly dependent on other cell type-specific factors (Fig. 4).

Open in a separate window

Fig. 4.

Sox17 transcriptionally regulates Wnt3A-mediated activation of both the Lef-1 promoter and TCF-responsive reporter TOP-flash. MCF-7, SW480, A549, or L cells were cotransfected with Wnt3A- (0.5 μg) and/or Sox17- (0.2 μg) expressing plasmid(s), together with the Lef-1 promoter luciferase (0.2 μg) (left) or TOP-flash luciferase (0.05 μg) (right) reporter. The pcDNA empty plasmid was used as a negative control and for normalizing total DNA content in each transfection. Each transfection cocktail also contained 0.01 μg of CMV-Renilla luciferase as an internal control plasmid for the normalization of transfection efficiency. At 24 h posttransfection, cell lysates were prepared, and both firefly and renilla luciferase activities were determined. Bars depict the mean (± SE, n = 3) RLU normalized for Renilla luciferase expression.

Sox17 repression of Wnt-mediated Lef-1 promoter induction requires the Sox17 HMG domain.

There are multiple mechanisms by which Sox17 can modulate Wnt activity. Sox17 can interact with β-catenin on DNA to regulate transcription or it can interact with and repress the activity of β-catenin/TCF/Lef-1 independently of DNA binding (28, 41, 42, 54). We sought to determine if Sox17 directly binds to and regulates the Lef-1 promoter. We performed structure-function studies using forms of Sox17 with mutations that effect specific domain functions (Fig. 5) [i.e., β-catenin binding domain (BBD) and high-mobility group (HMG) DNA binding domain]. We analyzed the ability of four mutant forms of Sox17 (Fig. 5, A and B) to interact with Lef-1 promoter sequences, TCF/Lef-1, and β-catenin, and for their ability to regulate the Lef-1 promoter. Only forms of Sox17 that could bind to DNA were able to regulate the Lef-1 promoter, suggesting a direct effect.

Open in a separate window

Fig. 5.

Structural analysis of Sox17-mediated regulation of the Lef-1 promoter by Wnt3A. Various deletion and point mutants of mouse Sox17 were used to determine which domains of Sox17 are required to repress Wnt3A-mediated transcriptional induction of the Lef-1 promoter. A: schematic representation of constructs encoding the full-length Sox17 protein and various Sox17 mutants (see materials and methods). All of the constructs contained a V5-tag at the COOH terminus. B: 293 cells were transfected with each of these constructs, and expression was evaluated by Western blotting with anti-V5 antibody at 24 h posttransfection. C: anti-V5 antibody was used to immunoprecipitate Sox17 from 293 cells transfected with the expression constructs shown in A, followed by Western blotting with anti-Sox17, anti-Lef-1, anti-TCF4, and anti-β-catenin antibodies. Nuclear extracts (500 μg) were used for IP reactions at 48 h posttransfection. The input lane represents direct loading of nuclear extract from full-length Sox17-transfected cells, whereas the No V5 lane omitted the capture V5 antibody from the IP reaction for full-length Sox17-transfected cells. D: analysis of the functions of different Sox17 domains in Wnt3A-mediated activation of Lef-1 promoter (top) and TCF-responsive reporter TOP-flash (bottom). 293 Cells were cotransfected with the Wnt3A (0.5 μg) expression plasmid and either the Lef-1 promoter (0.2 μg) or TOP-flash (0.05 μg) reporter plasmid, together with one of the indicated Sox17 (0.2 μg) expression plasmids. All transfections also contained 0.01 μg of a pCMV-Renilla luciferase plasmid for normalization of transfection efficiency. Bars depict the mean (± SE, n = 4) RLU at 24 h posttransfection and were normalized for Renilla luciferase expression.

As previously reported (41), full-length F-Sox17 associated with both endogenous β-catenin and TCF4 following immunoprecipitation of Sox17, and each of these mutants exhibited the predicted defects in associating with β-catenin or TCF4 (Fig. 5C). The TAD mutant lacked the ability to bind β-catenin but bound Lef-1 and TCF4, whereas the t-Sox17, Mut1, and Mut2 HMG binding domain mutants failed to bind Lef-1 or TCF4 but bound β-catenin (Fig. 5C). These findings substantiate previous results demonstrating that Sox17 can bind β-catenin and also associate with TCF4/Lef-1 through its HMG domain (41, 42, 54).

We next sought to evaluate whether Sox17/β-catenin and/or Sox17/TCF4 interactions are important for the ability of Sox17 to repress Wnt3A induction of the Lef-1 promoter in 293 cells. To distinguish between direct effects of Sox17 on the Lef-1 promoter and indirect effects on Wnt signaling, we also evaluated the Wnt/TCF-responsive TOP-flash reporter. One key difference was that the HMG domain of Sox17 differentially affected activation of the Lef-1 and Top-Flash promoters in the presence of Wnt3A (Fig. 5D). Wnt3A-mediated transcriptional activation of the Lef-1 promoter was not repressed by either the t-Sox17 or Mut1 protein, and repression by Mut2 was marginal compared with that effected by F-Sox17. By contrast, Sox17 Mut1 and Mut2 both significantly enhance TOP-flash activation in the presence of Wnt3A. Although Mut2 was previously shown to effect a similar enhancement of β-catenin-S37A-mediated TOP-flash in COS cells (41), Mut1 completely repressed β-catenin-S37A-mediated TOP-flash activation under the same conditions (41). Sox17/β-catenin interactions also contributed to repression of Wnt3A activation of both the Lef-1 and Top-flash promoters to varying extents, as the TAD mutant retained only partial activity compared with F-Sox17. Thus, context-dependent differences in the function of the Sox17 HMG binding domain appear to influence Wnt/β-catenin activation of both the Lef-1 and TOP-flash promoters.

Sox17 and TCF4 physically bind to the endogenous Lef-1 promoter in A549 cells.

We next performed ChIP assays to identify the regions of Sox17 and TCF4 binding to the Lef-1 promoter (Fig. 6). We utilized a lung carcinoma cell line (A549 cells) since this cell type is most relevant to our interest of Lef-1 regulation in airway epithelia. A549 cells were transfected with plasmids encoding V5-tagged-Sox17 or TCF4 before ChIP analysis. Immunoprecipitation was carried out using anti-V5 and anti-TCF4 primary antibodies (or nonimmune isotype-matched IgG alone as a negative control), and 10 oligonucleotide primer sets spanning the 2.5-kb Lef-1 promoter sequence were used for PCR amplification from these samples (Table 1 and Fig. 6A). Results from this analysis indicated that the endogenous Lef-1 promoter contains multiple Sox17 and TCF4 binding sites (Fig. 6, B and C). The identified binding regions overlapped with candidate consensus sequences for either Sox binding sites (SXS) or TCF/Lef-1 binding sites (TLS) (termed TLS1, TLS2, TLS3/4, SXS5, and TLS6; sites are numbered in sequence from the 5′ → 3' end of the promoter) (Fig. 6A). Quantitative PCR results demonstrated that Sox17 strongly bound to regions P2 (containing TLS2), P4 (containing SXS5), and P7 [containing TLS6 and the previously identified WRE (17)] of the Lef-1 promoter, with weaker binding at P1 (TLS1) (Fig. 6, B and C); notably, only one of these regions (P4) contained a Sox consensus site (SXS5). By contrast, TCF4 bound most strongly to the P3 region (TLS3/4) of the Lef-1 promoter and more weakly to the P1 (TLS1) and P7 (TLS6) regions. These finding demonstrate that both Sox17 and TCF4 can physically interact with the Lef-1 promoter.

Open in a separate window

Fig. 6.

ChIP survey of Sox17 and TCF4 binding sites in the endogenous Lef-1 promoter of A549 cells. A: schematic representation of the 10 PCR fragments used to survey the 2.5-kb human Lef-1 promoter (Table 3). Each PCR fragment covered 250–400 bp of the promoter. B and C: A549 cells were transfected with mouse Sox17- or TCF4-expressing plasmid, and genomic DNA sheared to an average size of 400–600 bp before ChIP was performed with anti-Sox17 and TCF4 antibodies. The recovered DNA was then evaluated by either standard PCR (B) or real-time PCR (C). B: agarose gel images show PCR products following 30 cycles of amplification from DNA immunoprecipitated with anti-Sox17 antibody (top), anti-TCF4 antibody (middle), or nonimmune total IgG from the same species as the primary capture antibody (bottom; used as a negative control). C: quantitative PCR results from the ChIP assays. Values depict the fold increase in copies of each target sequence detected with the specific capture antibody over the number detected with the species-matched IgG control antibody alone. Data are representative of 3 independent experiments.

Sox17 directly binds to sequences in the Lef-1 promoter.

Based on the overlap in ChIP binding of Sox17 to Lef-1 promoter regions that contain TCF consensus sites, we performed a more extensive sequence analysis of the Lef-1 promoter for TCF and Sox candidate binding sites. Six sites in the Lef-1 promoter (TLS1, 2, 3, 4, and 6) matched two widely accepted consensus sequences for TCF (CTTTGWW and CWTTGWW) (4, 38, 39) (Table 4). Of note, a widely accepted SRY/Sox consensus sequence (WWCAAWG) (22, 25, 49) is identical to the reverse-strand TCF consensus sequence CWTTGWW (Fig. 7A). Additionally, one core Sox consensus sequence (AACAAT) in the Lef-1 promoter (SXS5) did not conform to a TCF consensus sequence. Given that TCF and SRY/Sox proteins can bind to similar sequences with different affinities (22), this analysis suggested potential for context-specific binding of TCF4 and Sox17 to the same sites in the Lef-1 promoter (Table 4).

Open in a separate window

Fig. 7.

Evaluation of Sox17 and TCF binding sites in the Lef-1 promoter by EMSA. A: schematic representation of the 2.5-kb human Lef-1 promoter, with the positions and sequences of candidate TCF and Sox17 protein binding sites indicated (as determined based on the consensus sequences described in Table 4). The direction of arrows marked TCF and Sox17 indicate whether the consensus conforms to the forward or reverse stand of the DNA sequence. B: EMSA gels depicting protein/DNA complexes formed on the indicated biotinylated probes (Table 2) following binding by purified BSA (b), Sox17 (S), TCF4 (Tc), and Lef-1 (L). SP1 and TOP are negative and positive control biotinylated oligonucleotides for SP1 and TCF/Lef-1 (same consensus in TOP-flash) binding, respectively. C: EMSA gels depicting protein/DNA complexes formed when purified Sox17 protein was incubated with the indicated wild-type (W) or mutant (Mu) biotinylated probes (Table 2) for the 3 primary Sox17 binding sites. D: EMSA gels depicting protein/DNA complexes formed on the indicated biotinylated probes following binding of purified BSA (b), full-length Sox17 (S), Sox17-TAD (T), t-Sox17 (t), Sox17-mut1 (M1), or Sox17-mut2 (M2).

To test the potential molecular interactions at these sites in the Lef-1 promoter, we performed EMSA using recombinant His-tagged Sox17, TCF4, and Lef-1 proteins purified from bacteria. Results from these studies demonstrated that Sox17 binds strongly to double-stranded oligonucleotides (Table 2) containing TLS2, SXS5, and TLS6 sequences, and that it binds weakly to TLS1 and TLS3/4 (Fig. 7B). TCF4 and Lef-1 bound strongly to TLS3/4 and/or TLS6, but relatively weakly to TLS1. The TOP consensus sequence, which is known to avidly bind TCF proteins, strongly bound to both TCF4 and Lef-1, but not to Sox17 (Fig. 7B). A negative control SP1 consensus demonstrated no binding to Sox17, TCF4, or Lef-1 protein. These findings support the hypothesis that the Lef-1 promoter harbors overlapping sites of binding for Sox17 and TCF proteins and that the affinities of these proteins for these sites are sequence specific.

To confirm the specificity of Sox17 binding to these sites, we performed EMSAs with mutated DNA sequences for TLS2, SXS5, and TLS6 sites (Table 2), those that demonstrated the strongest binding by ChIP and EMSA. As shown in Fig. 7C, Sox17 effectively bound to the wild-type, but not mutated, sequences. We next evaluated which functional domains of Sox17 are required for efficient DNA binding of Sox17 at these sites. These studies utilized Sox17 mutant proteins shown in Fig. 5A. As anticipated, the HMG binding domain mutant (t-Sox17) was unable to bind to any of these three sites (Fig. 7D). Additionally, mutations in the HMG domain either significantly reduced, or complete ablated, Sox17 binding to the TLS2, SXS5, and TLS6 sites. Notably, the Sox17 TAD mutant (deletion of only the BBD) bound the SXS5 and TLS6 sites much more weakly than did the wild-type F-Sox17 protein, but bound the TLS2 site with normal affinity (Fig. 7D). These findings suggested that binding of Sox17 to SXS5 and TLS6 sites is influenced by its BBD. Collectively, the results from these EMSA experiments suggest that multiple distinct and overlapping Sox17 and TCF binding sites exist within the Lef-1 promoter and that the affinity of Sox17 for certain sites is influenced by its BBD.

Sox17 interacts with β-catenin, TCF4, and Lef-1 to form stable protein/DNA complex on Lef-1 promoter elements.

Given the potential for overlap in the binding of TCF4, Lef-1, and Sox17 to specific sites in the Lef-1 promoter, it was important to establish whether these proteins form intermolecular complexes while bound to Lef-1 promoter elements. Indeed, coimmunoprecipitation (co-IP) of V5-tagged-Sox17, TCF4, and β-catenin-S37A from nuclear extracts demonstrated that these three proteins interact to varying extents in 293 and SW480 cells (Fig. 8A). To determine how interactions between Sox17, TCF/Lef-1, and β-catenin influence binding at specific TCF/Sox sites (TLS2, TLS3/4, SXS5, and TLS6) in the Lef-1 promoter, we developed a biotin/avidin-based DNA binding assay in which biotinylated oligonucleotides are used to capture complexes from a mixture of purified recombinant Sox17, TCF4, Lef-1, and β-catenin proteins (Table 2, Fig. 8B). As predicted by EMSA data, only Sox17 was captured with TLS2 oligonucleotide from this mixture of proteins (Fig. 8C). Additionally, no β-catenin was associated with the Sox17 bound to the TLS2 sequence. By contrast, Sox17, TCF4, Lef-1, and β-catenin proteins all bound to the TLS3/4, SXS5, and TLS6 sequences. The fact that Sox17, TCF4, and Lef-1 were each able to bind to the TLS3/4 and TLS6 sequences as a mixture of proteins substantiated our results from the EMSA assays, which had indicated each of these proteins could individually bind these sites to various extents (Fig. 7). However, the affinity of Sox17 for the TLS6 and TLS3/4 sequences appeared to be either reduced or enhanced, respectively, by the presence of TCF4, Lef-1, and β-catenin. These findings suggest that TCFs can influence Sox17 binding to the Lef-1 promoter in a sequence-specific manner through either competition for the binding site or intermolecular interactions. Of greatest interest was the finding that TCF4 and Lef-1 were associated with the SXS5 site only in the presence of Sox17 (Fig. 8C); neither TCF4 nor Lef-1 was observed to bind SXS5 when they were incubated in isolation with the DNA (Fig. 7B). These findings suggest that Sox17 binding to SXS5 promotes the formation of complexes with TCF4/Lef-1 and β-catenin. Similar observations were made for the TLS3/4 site for which the presence of TCF4/Lef-1/β-catenin enhanced Sox17 binding (Fig. 8C) over that seen in the presence of Sox17 alone (Fig. 7B).

Open in a separate window

Fig. 8.

Sox17, TCF4, Lef-1, and β-catenin form intermolecular complexes on sequences within the Lef-1 promoter. A: the indicated cell lines were transfected with V5-tagged Sox17-, TCF4-, and β-catenin-expressing plasmids, and nuclear extracts were prepared at 48 h posttransfection. These extracts were then immunoprecipitated with antibodies against V5 (for Sox17), TCF4, or β-catenin (Bc) before being immunoblotted against the indicated antibodies. B: schematic representation of the strategy used in the experiments represented in C and D to detect protein/DNA complexes bound to biotinylated oligonucleotide sequences from the Lef-1 promoter. Purified recombinant proteins were prebound to the biotinylated oligonucleotides before they were captured on Avidin dynabeads and Western blotted with the appropriate antibodies. C: association of β-catenin, Sox17, TCF4, and Lef-1 with the indicated Lef-1 promoter sequences. An SP1 oligonucleotide served as negative control. D: association of β-catenin, TCF4, Lef-1, and Sox17 mutant (Sox17-TAD or t-Sox17) with the indicated Lef-1 promoter sequences.

To investigate whether certain Sox17 domains influence the formation of Sox17/TCF4/Lef-1/β-catenin complexes on the SXS5, TLS6, or TLS3/4 Lef-1 promoter sequences, we repeated the binding assays with the Sox17-TAD and t-Sox17 mutants. Interestingly, unlike full-length Sox17, Sox17-TAD was incapable of forming a complex on SXS5 with TCF4, Lef-1, or β-catenin (Fig. 8D vs. Fig. 8C), thus the Sox17 BBD domain is necessary for complex formation at SXS5. By contrast, enhanced Sox17 binding to the TLS3/4 site in the presence of TCF4/Lef-1/β-catenin did not require the Sox17 BBD (Fig. 8D). On the other hand, t-Sox17 lacking the HMG domain failed to bind DNA at any of the three sites analyzed in the presence of TCF4/Lef-1/β-catenin (Fig. 8D). The ability of Sox17-TAD, but not t-Sox17, to associate with TLS3/4 in the presence of TCF4/Lef-1/β-catenin (Fig. 8D) suggests that the interaction between the HMG domains of Sox17 and TCF4 stabilize the protein complex leading to increased DNA affinity.

Multiple Sox17 sites in the Lef-1 promoter control activation and repression in a β-catenin-dependent fashion.

To evaluate the functional importance of Sox17 and TCF binding sites in activation and repression of the Lef-1 promoter, we mutated these three binding sites (TLS2, SXS5, and TLS6) within the Lef-1 promoter-luciferase reporter, both individually and in combination, and assessed transcriptional activity in 293 cells expressing β-catenin-S37A or Sox17 (Fig. 9, A and B). Notably, mutation of each of these sites (TLS2, SXS5, TLS6) individually, and in one combination (TLS2/SXS5), significantly impaired β-catenin-mediated activation of the promoter. Although the baseline activity of the promoter was impaired by the triple mutation (3.6-fold), this mutant was significantly more responsive to β-catenin-S37A activation (6.2-fold) than the wild-type promoter (4.6-fold). We next evaluated which of the primary Sox17 binding sites were necessary for promoter induction in 293 cells by overexpressed Sox17 (Fig. 9B). We found that mutation of the TLS2 site significantly enhanced Sox17-mediated activation of the Lef-1 promoter, suggesting that this site may repress transcriptional activity when Sox17 is also bound to other sites in the promoter. In support of this hypothesis, the TLS2/TLS6/SXS5 mutant lost Sox17-dependent hyperactivation. The Sox17 BBD was also necessary for hyperactivation of the TLS2 mutant, since the Sox17-TAD mutant did not activate TLS2 mutant transcription above the wild-type promoter (Fig. 9B).

Open in a separate window

Fig. 9.

Transcriptional analysis of the Lef-1 promoter with mutated Sox17 and TCF4/Lef-1 binding sites. A: schematic representation of TCF/Lef-1 and Sox17 binding interactions on the Lef-1 promoter as determined by ChIP, EMSA, and DNA-affinity capture assays. The size of each arrowed box denotes the level of binding (i.e., smallest represents weakest binding). The presence of β-catenin (red) indicates that a particular factor is able to bind β-catenin and/or that β-catenin may have to interact with the BBD of Sox17 to mediate an intermolecular interaction (i.e., at SXS5). B: promoter activity when 293 cells were transfected with the indicated mutant or wild-type Lef-1 promoter-luciferase constructs (as indicated by color), together with plasmid expressing the protein(s) indicated on the x-axis. C: promoter activity when A549 cells were transfected with the indicated mutant or wild-type Lef-1 promoter-luciferase constructs on the x-axis, together with plasmid expressing Wnt3A and/or Sox17 (as indicated by color). Significant differences in expression levels as determined by Student's t-test (P < 0.05) are indicated by the connecting lines.

We next sought to evaluate which Lef-1 promoter mutations interfered with the ability of Sox17 to repress promoter activation by β-catenin-S37A. As previously observed (Fig. 3E), Sox17 expression led to significant inhibition of β-catenin-dependent activation of the wild-type Lef-1 promoter in 293 cells (Fig. 9B). By contrast, all seven Lef-1 promoter mutants lost Sox17-mediated repression in the presence β-catenin-S37A. These findings suggest that all Sox17 binding sites within the Lef-1 promoter are equally required for Sox17-mediated repression in the context of β-catenin-mediated activation. Also, promoter expression in the context of Sox17-TAD/β-catenin-S37A coexpression was indistinguishable from that produced by F-Sox17/β-catenin-S37A expression, regardless of which promoter construct (wild-type or any mutant) was tested. Thus, the BBD of Sox17 appears not to influence the transcriptional properties of Sox17 in the context of an activated β-catenin signal. In summary, these findings demonstrate that Sox17 can both activate and inhibit the Lef-1 promoter and that although the Sox17 BBD contributes to this activation, this region of the molecule is not required for promoter inhibition in the presence of a β-catenin signal.