Lef-1 isoforms regulate different target genes and reduce cellular adhesion

Abstract

The lymphoid enhancer factor 1 (Lef-1) belongs to the nuclear transducers of canonical Wnt-signalling in embryogenesis and cancer. Lef-1 acts, in cooperation with β-catenin, as a context-dependent transcriptional activator or repressor, thereby influencing multiple cellular functions such as proliferation, differentiation and migration. Here we report that an increased Lef-1 expression in human pancreatic cancer correlates with advanced tumour stages. In pancreatic tumours, two different transcripts of Lef-1 have been detected in various stages, as demonstrated by RT-PCR analysis. One transcript was identified as the full length Lef-1 (Lef-1 FL), whereas the second, shorter transcript lacked exon VI (Lef-1 Δexon VI) compared to the published sequence. Comparative analysis of these two Lef-1 variants revealed that they exhibit different cellular effects after transient expression in pancreatic carcinoma cells. Forced expression of Lef-1 Δexon VI inhibited E-cadherin expression in a β-catenin-independent way. Increased amounts of Lef-1 Δexon VI resulted in reduced cellular aggregation and increased cell migration. Expression of Lef-1 FL, but not the newly identified Lef-1 Δexon VI, induced the expression of the cell cycle regulating proteins c-myc and cyclin D1 in cooperation with β-catenin and it enhanced cell proliferation. Our findings indicate that expression of alternatively spliced Lef-1 isoforms is involved in the determination of proliferative or migratory characteristics of pancreatic carcinoma cells.

Lymphoid enhancer-binding factor 1 (Lef-1) and the structurally related T-cell transcription factors (TCF) belong to the family of high mobility group (HMG) proteins. Four members of the Lef/TCF-family have been described in vertebrates known as Lef-1, TCF-1, TCF-3 and TCF-4 which mediate canonical Wnt signalling during ontogenesis and in certain types of cancer, such as leukaemia, colon cancer or melanomas.1, 2, 3 Nuclear Lef/TCFs exhibit repressive activity towards the regulation of most target genes, whereas binding of β-catenin abolishes this repression thereby activates gene expression.4 Activation of the Wnt pathway results in inhibition of β-catenin degradation and recruitment of cytosolic β-catenin to Lef-1/DNA-binding sites.4 Inhibition of glycogen synthase kinase 3 activity plays an important role in this process.5 Lef/TCF-mediated Wnt signalling is involved in the regulation of a plethora of processes like cell proliferation, survival, differentiation or migration.

Lef/TCFs contain a highly conserved β-catenin-binding domain at the N-terminus. The highly conserved HMG-box localised at the C-terminus of Lef/TCF mediates sequence-specific DNA binding. The central core domain of Lef-1 and TCFs represents a context-dependent regulatory domain (CRD) and is specific for the individual Lef/TCFs. The importance of β-catenin-binding sites for its role in Wnt-signalling has been documented by the identification of an alternatively spliced Lef-1 isoform which lacks the N-terminal β-catenin binding site. This Lef-1 isoform represents a dominant negative Lef-1 due to prevention of β-catenin-recruitment to Wnt-responsive elements.3, 6

Up to now, more than 50 genes have been described to be regulated by the complex of β-catenin-Lef/TCF including those encoding for siamois, c-myc, c-jun, cyclin D1 or fibronectin7, 8, 9 (see also the Wnt-home page: www.stanford.edu/∼rnusse/wntwindow.html). The transcriptional activity of Lef-1 is further affected by interaction with additional regulatory factors, such as the repressor Groucho, the Smad proteins or Ets transcription factors, which partially enable Lef-1 to regulate gene expression independently of β-catenin.10, 11, 12

The individual members of the Lef/TCF family fulfil different roles in developmental processes. Lef-1 is expressed in developing thymus, involved in thymocyte differentiation as well as development of hair follicles, teeth and the mammary gland.13 In the adult organism, Lef-1 is essential for maintenance of the stem cell population in crypts of the intestinal tract.5 Cell culture experiments demonstrate a role of Lef-1 in induction of epithelial cell polarisation. TCF-4 has a specific role in maintenance of the intestine,14, 15 whereas TCF-3 has non-redundant functions in early embryonic development.16

With regard to the involvement in tumourigenesis, activated canonical Wnt-signalling has been reported in many tumours such as malignant melanoma and gastrointestinal malignancies, such as colorectal and hepatocellular cancer.13 Especially in the development of colon cancer, the role of TCF-signalling has been extensively studied.5, 17 The function of Lef-1 in induction and development of pancreatic ductal adenocarcinomas (PDAC) has not been analysed so far.

In the present study, we report the characterisation of a new Lef-1 isoform in pancreatic cancer, which lacks exon VI (Lef-1 Δexon VI) in comparison to the published Lef-1 sequence. In cell culture experiments, expression of Lef-1 Δexon VI specifically enhanced cell migration of pancreatic cancer cells which suggests that its expression during tumourigenesis may promote metastasis of pancreatic carcinoma.

ChIP: chromatin immunoprecipitation; CRD: context-dependent regulatory domain; dn: dominant negative; EGFP: enhanced green fluorescence protein; FL: full length; HMG: high mobility group; Lef-1: lymphoid enhancer factor 1; MPO-1: myeloperoxidase 1; PanIN: pancreatic intraepithelial neoplasia; PDAC: pancreatic ductal adenocarcinoma; RPLP0: ribosomal protein large P0; RPL30: ribosomal protein L30; TCF: T-cell transcription factor

Material and Methods

Tumour material

The departments of Surgery at the Universities of Ulm and Bochum (Germany) provided samples of ductal pancreatic adenocarcinoma obtained by surgical resections. Tissues of PanIN lesions were obtained from the Department of Pathology, University of Ulm (Germany). Normal pancreatic samples were gained from healthy areas out of pancreatic samples. Informed consent was given by all patients prior to usage of tissue or biopsy samples. The study was approved by the local ethics committees of the Universities of Ulm and Bochum.

Antibodies and plasmids

Antibody against β-actin (A-5441) was obtained from Sigma-Aldrich (Taufkirchen, Germany), against E-cadherin (#610182) and c-myc (#554205) from BD Biosciences (San Diego, CA), against fibronectin (#A103) and cyclin D1 (#06-137) from Biomol (Hamburg, Germany), against GFP (#1814460) from Roche Diagnostics (Mannheim, Germany) and HRP-coupled anti-mouse or anti-rabbit antisera from Pierce (#A1430 and #31460, Rockford, IL). Antibodies against p300 (sc-584) were purchased from Santa Cruz (Santa Cruz, CA, USA), against Lef-1 for CHIP assay from CST (#2230).

Lef-1 FL and Lef-1 Δexon VI were isolated by PCR using pfx polymerase (Gibco-Invitrogen, Karlsruhe, Germany), specific primer (sense: ATGCCCCAACTCTCCGGAGG, antisense: TCAGATGTAGGCAGCTGTCATTC) and cDNA from pancreatic cancer samples (No. 85 and 183) and from the colorectal cancer cell line SW480 as template. The obtained PCR fragments were cloned into pcDNA3.1 (Invitrogen, Karlsruhe, Germany) and pEGFP-C2 expression vector (BD Clontech, Heidelberg, Germany) using the EcoR1 restriction site. Lef-1 constructs devoid of the N-terminal β-catenin-binding domain were obtained by deleting the first 324 bp by EcoR1 treatment of pEGFP/Lef-1 plasmids and subsequent relegation. Correct orientation and reading frames were confirmed by sequencing (GATC Biotech, Konstanz, Germany).

Luciferase reporter constructs are listed in Supporting Information Table 1. A reporter for fibronectin was provided by D. Gradl (University of Karlsruhe, Germany),9 for E-cadherin by W. Birchmeier (Max-Delbrück-Center Berlin, Germany)18 and M. Park (University of Montreal, Canada), for cyclin D1 by R. Schmid (Techn. University of Munich, Germany)19 and a set of c-myc promoter constructs by B. Kinzler (John Hopkins University, Baltimore, USA).7, 20 The later contructs comprise a sequence of 1237 base pairs 5′ of the c-myc transcription start as well as promoter sequences containing base pair -783 and -26 bp to the start codon.

Cell culture

The pancreatic cancer cell lines PANC-1 (American Type Culture Collection, Rockville, MD; CRL-1469) and PaTU8902, PaTU8988s and PaTU8988t (German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany, ACC 179, ACC 204 and ACC 162) were maintained in DMEM supplemented with 10% FCS (Gibco-Invitrogen, Karlsruhe, Germany). For transient transfection of PANC-1, PaTU8902 cells and PaTU8988t 2 × 10⁵ cells were seeded in 60 mm dishes. After 12 h cells were transfected at 70% density with plasmids or siRNA as indicated using DMRIE-C (Invitrogen) as transfection reagent according to manufactures instruction. Transfections with efficiencies below 35–40% were excluded from further experiments.

RT-PCR studies

RNA was extracted using the RNeasy Mini Kit (Qiagen, Hilden, Germany). To produce cDNA, 1 μg of total mRNA was transcribed using random primers and Superscript II reverse transcriptase (Invitrogen). The semi-quantitative PCR was performed with specific primers for Lef-1, fibronectin, E-cadherin, cyclin D1, c-myc and β-actin. The specificity of primer pairs was verified by sequencing RT-PCR products.

Quantitative real-time PCR was performed using the comparative C_T method, Lef-1-specific primers (sense: AATGCACGTGAAGCCT; antisense GAATCTGGTTGATAGCTGC) and SYBR Green PCR Master Mix (Applied Biosystems) with an ABI Prism 7700™ Sequence Detection System (Applied Biosystems, Foster City, CA) using according to the manufacture's instructions. Sequence-specific primer pairs were determined with the PrimerExpress 2.0 program (Applied Biosystems). The relative gene expression in relation to ribosomal protein RPLP0 expression (sense: AGATCCGCATGTCCCTTC; antisense CCTTGCGCATCATGGTGTT) is given as means ± SEM performed in duplicate using three independent cDNAs.

Western blotting

Protein analyses were performed as previously described by Seidel et al.21 Briefly, cells were lysed using radioimmunoprecipitation assay buffer (1% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodececylsulfate, 150 mmol/L NaCl, 50 mmol/L Tris/HCl (pH 7.2), 10 mmol/L EDTA, 10 mmol/L EGTA) containing 400 μmol/L aprotinin, 50 μmol/L leupeptin, and 0.5 mmol/L Pefabloc (all from Roche Diagnostics) to inhibit proteases. Thirty microgram of total lysates were analysed by SDS-PAGE, blotted onto nitrocellulose and incubated with the indicated antibody over night and a HRP-labelled secondary antibody. Immunoreactive proteins were visualized using an enhanced chemiluminescence detection system (Pierce).

ChIP assay

The binding of Lef-1 to the E-cadherin promoter was analysed by chromatin-immunoprecipitation assay (ChIP) using the ChIP enzymatic chromatin IP kit from CST (#9003) according to manufactures instruction. DNA was digested using Micrococcal nuclease (3000 units, CST). Immunoprecipitations were performed with a Lef-1-specific (#2230; CST) or a GFP-specific (#600-301-215; Rockland Immunochemicals) antibody and lysate of PANC-1 or PaTU8902 cells transiently expressing EGFP, EGFP-Lef-1 FL or EGFP-Lef-1 Δexon VI. Lef-1-binding to the E-cadherin promoter was examined by real-time PCR using specific primer pairs according to published sequence of CDH1 (NG_008021)22 (Ecad1400: sense ACTGCAAGCTCCACCT; antisense GGTCAGGAGATTGAGACCA; Ecad400: sense TGTGATCG CACCACTG; antisense CGGTTCTGATTCCACTG; Ecad200: sense CTAGCAACTCCAGGCT; antisense TGCTTTGCAG TTCCGA). PCRs were performed in duplicate. As positive control histone H3 antibody (#2650, CST) was used for precipitation and as negative control normal rabbit IgG (#2729, CST) was used. ChIP assays were performed with two unrelated cell lysates.

Proliferation, aggregation and migration assay

To determine the cell-cell adhesion capacity, cell aggregation assays were performed as described in Vogelmann et al.23 Cell adhesion was analysed 36 h after transfection with pEGFP/Lef-1 ΔexonVI, pEGFP/Lef-1 FL or pEGFP in rotation aggregation assay. The aggregation index was calculated from the ratio of the aggregate number after 30 min of rotation and the particle number at the beginning (A_i = (N₀–N₃₀)/N₀). Five independent assays were performed in duplicate.

Migration and invasion assays were performed as described in Vogelmann et al.23 The number of cells that had migrated through uncoated or collagen type I-coated transwell inserts (8,0 μm pore size, BD Bioscience) was estimated 36 h after transfection of PANC-1 or PaTU8902 cells with pEGFP/Lef-1 ΔexonVI, pEGFP/Lef-1 FL or pEGFP by counting three independent, randomly chosen visual fields by phase contrast microscopy. Three independent assays were performed in duplicate.

Cell proliferation was measured by counting total cell number using the cell counter CASY TTC (Schaerfe System, Reutlingen, Germany). Five × 10⁴ cells were seeded in 60 mm cell culture dishes and transfected after 18h with pEGFP/Lef-1 ΔexonVI, pEGFP/Lef-1 FL or pEGFP. Transfections with efficiencies below 80% were excluded from the study. On subsequent days, the adherent cells were harvested, resuspended in PBS and the cell number was counted using two different cell dilutions as described in Voisard et al.24 Cell numbers ± SEM per 60 mm dish were given in relation to the number of seeded cells from three independent experiments. For statistical analysis, Student's t-test was used and p < 0.05 was considered significant.

For quantification of mitotic figures, cells were cultured on coverslips and transfected with pEGFP/Lef-1 ΔexonVI, pEGFP/Lef-1 FL or pEGFP using DIMRIE-C as transfection reagent. After 48 h cells were fixed with ice-cold methanol/acetone (1:1) and stained with 4,6-diamidino-2-phenylindol (DAPI; 0.1 μg/mL). Only mitotic figures of cells showing an EGFP signal were counted and the proliferation indices were calculated from 10 fields (120 × 120 μm) per coverslip of three independent experiments.

Gene reporter assay

At 60% of confluence, PANC-1 and PaTU8902 cells were co-transfected with constructs containing fragments of the fibronectin, E-cadherin, c-myc or cyclin D1 promoter and a pRLTK vector (Promega, Mannheim, Germany) containing the renilla firefly gene under the control of a thymidine kinase promoter. The cells were harvested 36 h after transfection and processed using the Dual Luciferase Kit (Promega) as described by the manufacturer. Luciferase activity was normalised to renilla firefly activity.

For statistical analysis, Student's t-test was performed and p < 0.05 was considered significant.

Results

To elucidate the role of Lef-1 in the development of pancreatic cancer, we analysed the expression of Lef-1 in pancreatic tumour samples obtained from surgical resections. RT-PCR studies revealed Lef-1 expression in 11 out of 15 different pancreatic cancer samples but only small amounts of Lef-1 were detected in normal pancreatic tissue (0/4) (Fig. 1a). Lef-1 is predominantly expressed in more advanced tumour stages (7/9; pathological staging: pT3-4 N1 Mx) while a moderate expression was detected in early tumour stages (3/6; pathological staging: pT1-2 N0 M0). No correlation was observed between expression of Lef-1 and of myeloperoxidase 1 (MPO-1), a lymphocyte marker protein (Fig. 1a). This lack of correlation implicates that the observed Lef-1 signal was not contributed by infiltrating lymphocytes but by in the tumour cells themselves. The expression of Lef-1 in pancreatic cancer was underlined by quantitative RT-PCRs using microdissected tissue samples of different pancreatic tumour stages (pancreatic intraepithelial neoplasia (PanINs)25 and invasive pancreatic ductal adenocarcinoma (PDAC) of four different patients compared with samples of normal pancreatic ducts. Figure 1b demonstrates an enhanced Lef-1 expression in PanIN stage III and a nearly fourfold increase in invasive cancer samples compared to control tissues. cDNA from SW480 cells served as positive control.

Figure 1. Lef-1 expression in pancreatic cancer. (a) cDNA from different human pancreatic tissue samples was analysed by PCR for the presence of transcripts encoding for Lef-1, myeloperoxidase-1 (MPO-1) and β-actin. Samples from advanced tumours are grouped into the pathological stages pT3-4 N1 Mx and compared with early tumour stages of pT1-2 N0 M0 as well as samples of normal pancreatic tissue. The colorectal cancer cell line SW480 which expresses high amounts of Lef-1, served as positive control. PCR with β-actin-specific primers was used to document equal amounts of cDNA used in the reaction. (b) The relative amount of Lef-1 mRNA was estimated in defined tumour stages of different pancreatic intraepithelial neoplasia (PanIN-stages) and invasive pancreatic ductal adenocarcinoma isolated by microdissection and analysed by quantitative RT-PCR. The relative Lef-1 expression was normalised to the expression of the ribosomal protein RPLP0. At least four different samples were used for each PanIN-stage. Means ± SEM are shown and * represents statistically significant differences (Student's t test, p < 0.05 was considered significant). (c) Detection of Lef-1 mRNA from cDNAs of different human pancreatic carcinoma cell lines by semi-quantitative PCR using Lef-1-specific primers. Amplification of β-actin was used to document equal amounts of cDNA. A representative gel out of three independent experiments is shown.

The RT-PCR studies revealed that most of the analysed pancreatic tumours displayed two Lef-1 bands (Fig. 1a). To understand the nature of these double bands, complete sequences of both Lef-1 cDNAs were determined. The longer variant corresponds to the sequence published for full length Lef-1 (Lef-1 FL, Acc.No. AF288571, NM_016269). The shorter variant missed 84 base pairs, corresponding to the entire exon VI of the Lef-1 sequence and coding for a serine rich part of the Lef-1 context-dependent regulatory domain (Supporting Information Fig. 2c). This Lef-1 variant was called Lef-1 Δexon VI. Analyses of pancreatic carcinoma cell lines verified the existence of two Lef-1 transcripts in half of the analysed cell lines. PaTU8988t and Capan-2 expressed both isoforms at high level, whereas AsPC-1 expressed only the Lef-1 Δexon VI isoform (Fig. 1c).

Lef-1 Δexon VI regulates gene expression of fibronectin and E-cadherin

To compare the cellular function of both Lef-1 variants, we expressed the isoforms as EGFP-fusion proteins in the pancreatic carcinoma cell lines PANC-1 und PaTU8902, which expressed no or only marginal levels of Lef-1 as analysed by RT-PCR (Fig. 1c).

PANC-1 or PaTU8902 cells expressing EGFP/Lef-1 Δexon VI did not exhibit striking alterations in their cell morphology in comparison to EGFP/Lef-1 FL-expressing cells or EGFP-expressing controls (data not shown).

To determine the influence of Lef-1 Δexon VI on transcriptional regulation, the expression of known Lef-1/TCF target genes was analysed by RT-PCR, luciferase reporter assays and Western blot analyses.

Expression of EGFP/Lef-1 Δexon VI resulted in enhanced amounts of fibronectin mRNA and protein compared to expression of EGFP/Lef-1 FL or EGFP in PANC-1 and PaTU8902 cells (Fig. 2a, 2c). RT-PCR experiments showed a threefold increase in fibronectin mRNA induced by expression of EGFP/Lef-1 Δexon VI (Fig. 2a). This increase was due to enhanced transcriptional activity as demonstrated by luciferase gene reporter assays. Using a fibronectin promoter fragment, luciferase activity was 3.2-fold higher in cells cotransfected with pEGFP/Lef-1 Δexon VI compared to PANC-1 cells cotransfected with pEGFP/Lef-1 FL or pEGFP (Fig. 2b). Similar results were obtained for PaTU8902 cells.

Figure 2. Expression of EGFP/Lef-1 Δexon VI in PaTU8902 cells reduces E-cadherin and increases fibronectin expression. (a) The amount of fibronectin and E-cadherin mRNAs was analysed by semi-quantitative RT-PCR in PaTU8902 cells transiently transfected with pEGFP, pEGFP/Lef-1 FL or pEGFP/Lef-1 Δexon VI 36 h after transfection. Lef-1 was amplified as internal control and β-actin served to document equal amounts of cDNA. (b) The activity of the fibronectin promoter was studied in cells cotransfected with pEGFP, pEGFP/Lef-1 FL, pEGFP/Lef-1 Δexon VI, pEGFP/Lef-1 Δexon VI ΔHMG or pEGFP/Lef-1 ΔNΔexon VI and reporter constructs containing a fragment of the fibronectin promoter (−499/+20). (c) Western blot analyses of fibronectin, E-cadherin and EGFP in total lysates of transfected PANC-1 cells are shown. Staining of β-actin documents the use of equal amounts of protein in each lane. (d) The activity of the E-cadherin promoter-specific reporter construct was analysed in PANC-1 and PaTU8902 cells cotransfected with pEGFP, pEGFP/Lef-1 FL, pEGFP/Lef-1 Δexon VI, pEGFP/Lef-1 Δexon VI ΔHMG or pEGFP/Lef-1 ΔNΔexon VI, respectively, and a mouse E-cadherin promoter-containing reporter (−306/−36). The indicated sample (+ Sb) was treated with 1 μM TGFβ inhibitor Sb505124. (e) The activity of a mutated E-cadherin promoter construct containing a non-functional E-box2 was analysed in PANC-1 cells cotransfected with pEGFP, pEGFP/Lef-1 FL, pEGFP/Lef-1 Δexon VI and the mutated E-cadherin promoter. Luciferase activity was normalised to renilla activity and expressed in relation to fibronectin or E-cadherin promoter activity in pEGFP-transfected controls. Means ± SEM of three independent experiments performed in duplicate are shown. (f) Inhibition of Lef-1 expression by siRNA in PaTU8988t cells was performed by siRNAs homolog to the C-terminus of Lef-1 or homolog to exon VI. Non-specific siRNA served as control. Successful Lef-1 repression was analysed by Western blotting. Protein concentration of the Lef-1 targets E-cadherin, cyclin D1 and c-myc were determined by Western blotting. Detection of β-actin served as control for equal amounts of protein used in each lane.

To analyse whether the Lef-1 Δexon VI-induced activity of the fibronectin promoter depends on the ability of Lef-1 Δexon VI to bind to DNA, we expressed a mutated construct of Lef-1 Δexon VI lacking the HMG-box. The HMG-box is responsible for the sequence-specific DNA-binding of Lef-1, and its deletion results in a dominant negative variant of Lef-1 (dnLef-1).6 In contrast to EGFP/Lef-1 Δexon VI, expression of EGFP/Lef-1 Δexon VI ΔHMG did not activate the fibronectin promoter in the luciferase gene reporter assay (Fig. 2b). These data are in agreement with results obtained with cells of the colorectal cancer cell line SW480, which contain both variants of Lef-1. Suppression of Lef/TCF signalling in SW480 cells by means of dnLef-1 Δexon VI ΔHMG-expression decreased fibronectin promoter activity by 57% (data not shown).

Furthermore, we examined whether the effect of Lef-1 Δexon VI on the fibronectin promoter depends on its interaction with β-catenin. An EGFP/Lef-1 Δexon VI mutant, which lacks the N-terminal β-catenin binding site, (EGFP/Lef-1 ΔNΔexon VI) was cotransfected with the fibronectin reporter construct. Luciferase assays revealed that Lef-1 ΔNΔexon VI was no longer effective in the regulation of fibronectin gene expression (Fig. 2b) suggesting that Lef-1 Δexon VI activates the fibronectin promoter after binding of β-catenin.

In contrast to the stimulatory effect of Lef-1 Δexon VI on fibronectin expression, Lef-1 Δexon VI inhibited E-cadherin expression on mRNA as well as protein level in PANC-1 and PaTU8902 cells (Fig 2a, 2c). Figure 2c shows that the protein level of E-cadherin was clearly reduced in EGFP/Lef-1 Δexon VI-expressing cells. Luciferase gene reporter assays using a mouse E-cadherin promoter fragment confers that EGFP/Lef-1 Δexon VI-expression inhibited the activity of the E-cadherin promoter in PANC-1 and PaTU8902 cells by 50% and 46%, respectively, compared to mock transfected controls or EGFP/Lef-1 FL-expressing cells (Fig. 2d). Expression of the dominant negative Lef-1 Δexon VI ΔHMG construct did not affect E-cadherin promoter activity (Fig. 2d and Supporting Information Fig. 1). Most interestingly, the Lef-1 ΔNΔexon VI mutant, which lacks the β-catenin binding site, repressed E-cadherin gene expression pointing to an β-catenin-independent effect of regulating E-cadherin gene expression. Recently, an interaction between Smad transcription factors and Lef/Tcf has been suggested in the E-cadherin regulation.26 In the cell system used in this study, the inhibition of the TGFβ₁ receptor type I kinase by a pharmacological inhibitor (Sb505124) resulted in a clear suppression of EGFP/Lef-1 Δexon VI-induced repression of the E-cadherin promoter (Fig. 2d). It has been suggested that two E-boxes which were characterised in the mouse and the human E-cadherin promoter ((148 (E-box1) and 201 (E-box2) base pairs upstream of the start codon) play a central role in the regulation of E-cadherin expression.18, 26–28 Using a mutated E-cadherin promoter construct, comprising a non-functional E-box2 sequence (Supporting Information Table 1), neither Lef-1 FL nor Lef-1 Δexon VI altered the promoter activity (Fig. 2e). These data strongly suggest that the regulation of the E-cadherin promoter, but not of the fibronectin promoter, by Lef-1 Δexon VI is independent of Lef-1/β-catenin binding.

Lef-1 knock-down increased E-cadherin concentration in pancreatic carcinoma cells

The pancreatic carcinoma cell line PaTU8988t exhibits both Lef-1 isoforms as shown by RT-PCR in Figure 1c. To confirm the connection between Lef-1 Δexon VI and E-cadherin expression, we inhibited Lef-1 expression in PaTU8988t cells by siRNA treatment. Transfection of Lef-1 siRNAs resulted in reduced expression of both Lef-1 isoforms (Fig. 2f) even when siRNAs were used which showed homology only for the sequence of exon VI of Lef-1 (supporting Information Table 1). The knockdown of Lef-1 resulted in increased E-cadherin protein concentrations in PaTU8988t cells (Fig. 2f) which is in agreement with the influence of Lef-1 Δexon VI on the E-cadherin promoter shown before.

Lef-1 Δexon VI binds to the E-cadherin promoter

Binding of the Lef-1 isoforms to the human E-cadherin promoter was further analysed by chromatin IP assays using PCR primer specific for published Lef-1-binding sites of the E-cadherin promoter.26, 22 To avoid non-specific precipitates of the Lef-1 antibody or the influence of endogenous Lef-1, the fusion protein of EGFP-Lef-1 expressed in PANC-1 and PaTU8902 cells was precipitated using a GFP-specific antibody. Real time PCR data show that a DNA fragment which was amplified using primer spanning a region of about 100 to 270 base pairs 5′ of the E-cadherin start codon including the classical E-box1+2 (Ecad200)29 was markedly precipitated with EGFP-Lef-1 Δexon VI (Fig. 3b). A further upstream localised potential Lef-1 binding site (Ecad400),26 covered by a sequence between 337 and 514 bp upstream of the start codon, was rarely coprecipitated together with EGFP-Lef-1 Δexon VI (Fig. 3a). A DNA-fragment nearly 1 kb upstream of the E-cadherin start codon was not amplified from precipitates of Lef-1 or EGFP-Lef-1 (Fig. 3c). These data confirm that primarily Lef-1 Δexon VI was associated with the E-cadherin promoter in contrast to EGFP-Lef-1 FL or EGFP (Fig. 3). Lef-1 Δexon VI bound mainly to or near the classical E-box1+2, located 200 bp upstream of the start codon of E-cadherin. Whereas only a weak PCR signal was detected after amplification of a Lef-1 element located 400 bp upstream of the human E-cadherin start codon26 (Fig. 3a). This interaction was independent of the cell line (PANC-1 or PaTU8902) used for the assays. Precipitation of EGFP/Lef-1 FL yielded comparable amounts of PCR products as precipitation of the EGFP-protein alone.

Figure 3. Lef-1 Δexon VI binds to the human E-cadherin promoter. Chromatin immunoprecipitation assays were performed to identify Lef-1 proteins associated with the human E-cadherin promoter. After precipitation of Lef-1 or GFP using specific antibodies, associated DNA was identified by real time PCR using primer pairs specific for two potential Lef-1-binding sites (Ecad400 and Ecad200) of the E-cadherin promoter and compared to a DNA sequence 1 kb upstream (Ecad1400). Two independent chromatin-protein precipitates were analysed by real time PCR.

Expression of Lef-1 FL activates c-myc and cyclin D1 expression

In the following studies, we examined whether additional Lef-1 target genes were also differentially regulated by the Lef-1 isoforms. In PANC-1 and PaTU8902 cells expressing EGFP/Lef-1 FL we observed increased amounts of c-myc and cyclin D1 mRNAs, whereas expression of EGFP/Lef-1 Δexon VI did not change these mRNA concentrations as analysed by semi-quantitative RT-PCR (Fig. 4a). To confirm the observed effects, we determined the influence of both Lef-1 isoforms on the promoter activity of c-myc and cyclin D1 genes using specific luciferase gene reporter assays. Figure 4c shows that only expression of EGFP/Lef-1 FL significantly increased the activities of the c-myc and the cyclin D1 luciferase reporter in contrast to cells expressing EGFP/Lef-1 Δexon VI or EGFP. An increase of c-myc and cyclin D1 was also confirmed on protein level by Western blotting procedure. Figure 4b documents that PaTU8902 cells expressing EGFP/Lef1 FL exhibit a marked increase of c-myc and cyclin D1 content in total protein lysates compared to EGFP- or to EGFP/Lef-1 Δexon VI-expressing PaTU8902 cells. Similar results were obtained using PANC-1 cells (data not shown).

Figure 4. EGFP/Lef-1 FL induces enhanced expression of the cell cycle regulators c-myc and cyclin D1. (a) The amount of cyclin D1 and c-myc mRNA was analysed by semi-quantitative RT-PCR in PaTU8902 cells transiently transfected with pEGFP, pEGFP/Lef-1 FL or pEGFP/Lef-1 Δexon VI 36 h after transfection. Lef-1 was amplified as internal control and β-actin served to document equal amounts of cDNA. (b) Western blot analyses of cyclin D1, c-myc and EGFP were performed using total lysates of transfected PaTU8902 cells. Staining of β-actin documents equal amounts of protein in each lane. (c) The promoter activity of the cyclin D1 and c-myc genes was determined in PaTU8902 cells cotransfected with pEGFP, pEGFP/Lef-1 FL or pEGFP/Lef-1 Δexon VI and a luciferase reporter. The reporter constructs contained cyclin D1 (−673/+135) or c-myc promoter (−1237/+3) fragments. Luciferase activity was normalised to the activity of a cotransfected renilla construct and was expressed in relation to the promoter activity in pEGFP-transfected controls. Means ± SEM of three independent experiments performed in duplicate are shown. (d) Inhibition of β-catenin expression by siRNAs in PaTU8988t cells. Successful β-catenin repression is demonstrated by Western blotting. Protein concentration of the Lef-1 target c-myc was analysed by Western blotting. Detection of β-actin served as control for equal amounts of protein used in each lane.

As described before for Lef-1-induced E-cadherin repression, the inhibition of both Lef-1 isoforms by siRNA treatment reverted the elevated cyclin D1 and c-myc protein concentrations in PaTU8988t cells (Fig. 2f). In contrast to the β-catenin-independent repression of E-cadherin by Lef-1 Δexon VI, the effect of Lef-1 FL on the c-myc promoter strongly depends on the presence of β-catenin. A siRNA-mediated knock-down of β-catenin expression repressed the increase of c-myc in PANC-1 cells co-transfected with EGFP/Lef-1 FL on protein level (Fig. 4d).

Lef-1 FL binds to the Lef/TCF-binding site of the c-myc promoter

To analyse the interaction of the Lef-1 isoforms with the c-myc promoter, we used differently sized constructs of the c-myc reporter as described before.7, 20 Comparing the different Lef-1 isoforms Lef-1 FL but not Lef-1 Δexon VI or the mutants missing the β-catenin binding site (Lef-1 ΔN FL or Lef-1 ΔNΔexon VI) increased the activity of the c-myc reporter in PaTU8902 cells (Fig. 5a).

Figure 5. EGFP-Lef-1 FL activates the c-myc promoter by β-catenin-dependent binding to two Lef-1-binding sites. (a) The activity of human c-myc promoter fragments was analysed by luciferase reporter assays in PaTU8902 cells. Cells were cotransfected with pEGFP, pEGFP/Lef-1 FL, pEGFP/Lef-1 Δexon VI, pEGFP/Lef-1 ΔN FL or pEGFP/Lef-1 ΔNΔexon VI and c-myc promoter constructs containing base pairs −1237 to +3 (pGl/1237), −783 to +3 (pGl/783), −26 to +3 (pGl/26). Luciferase activity was normalised to the activity of a cotransfected renilla construct and expressed as x-fold increase in relation to activity of the empty pGL3 luciferase vector. Means ± SEM of two independent experiments performed in duplicate are shown. (b) MMP7 and NrCAM mRNA concentrations were analysed by semiquantitative RT-PCR using PaTU8902 cells transiently transfected with pEGFP, pEGFP/Lef-1 FL or pEGFP/Lef-1 Δexon VI. Lef-1 was amplified as internal control and amplification of β-actin served to document equal amounts of cDNA.

The c-myc promoter construct pGl/myc1273 (−1273/+3) which contains two Lef-1-binding sites (TBE1 and TBE2)7 showed the highest c-myc promoter activity after cotransfection with Lef-1 FL (Fig. 5a). In contrast, Lef-1 FL increased the activity of the pGl/myc783 construct (−783/+3), which includes only the TBE2 site,7 nearly half compared to the pGl/myc1273 promoter (Fig. 5a) suggesting that Lef-1 FL binds to both Lef/TCF-binding sites in the c-myc promoter. The activity of both reporter constructs was not altered by coexpression of Lef-1 Δexon VI (Fig. 5a). The construct pGl/myc26 (−26/+3), which contained a minimal promoter sequence without Lef-1-binding sites, did not induce an increase in promoter activity by one of the Lef-1 isoforms.

In contrast to these differential effects of the two Lef-1 isoforms, both Lef-1 variants modulated the expression of another set of known Lef/TCF target genes, namely NrCAM,30 MMP731 and claudin-132 (Fig. 5b, Supporting Information Fig. 1b). The mRNA amount of NrCAM was reduced and the amounts of MMP-7 and claudin-1 were elevated in pancreatic carcinoma cells transfected with pEGFP/Lef-1 Δexon VI or pEGFP/Lef-1 FL compared to mock transfected controls (Fig. 5b and Supporting Information Fig. 1b).

Lef-1 Δexon VI reduces cell aggregation and increases cell migration

E-cadherin represents the main cell-cell adhesion protein in epithelial cells and its altered concentration influences cellular adhesion as well as cell migration.33 To determine the correlation of Lef-1 transcriptional activity with cellular migration and adhesion, we performed aggregation and migration assays using Lef-1 Δexon VI- or Lef-1 FL-expressing pancreatic carcinoma cells. Cell-cell aggregation assays demonstrated a reduced aggregation capacity of PANC-1 cells expressing EGFP/Lef-1 Δexon VI compared to cells with EGFP/Lef-1 FL or mock transfected controls (Fig. 6a). The observed inhibition was comparable to the one obtained after inhibition of E-cadherin by the addition of a neutralising E-cadherin antibody to EGFP/Lef-1 FL-expressing cells (Fig. 6a).

Figure 6. Cell proliferation, cell-cell aggregation and invasion rates of PaTU8902 cells transfected with pEGFP, pEGFP/Lef-1 FL or pEGFP/Lef-1 Δexon VI. (a) The cellular aggregation capacity was estimated by rotation aggregation assays of transiently transfected PANC-1 cells. The addition of a neutralising antibody against E-cadherin (anti-E-cad) verified the contribution of E-cadherin to the observed aggregation. Cells were singularised 36 h after transfection and aggregation was measured under constant rotation of 80 rpm. The aggregation index was determined by A_i = (N₀–N₃₀)/N₀, with N₀ representing particle number before and N₃₀ after 30 min of aggregation. Means ± SEM of five independent experiments performed in duplicate are given. (b) Migration of PANC-1 cells expressing EGFP, EGFP/Lef-1 FL or EGFP/Lef-1 Δexon VI was measured using non-coated transwell migration chambers. The number of cells which had migrated through the porous membrane was estimated in three visual fields by phase contrast microscopy and is given as means ± SEM of three different assays performed in duplicate. (c) Cell invasion through a collagen gel was measured using transwell migration chambers with a collagen type I-coated porous membrane. Cells were transfected with pEGFP, pEGFP/Lef-1 FL or pEGFP/Lef-1 Δexon VI. 48 hours after transfection, the number of cells which had migrated through the collagen-coated membrane was estimated microscopically in three visual fields. Means ± SEM of three independent assays performed in duplicate are shown. (d) Proliferation of PANC-1 cells expressing EGFP, EGFP/Lef-1 FL or EGFP/Lef-1 Δexon VI was analysed estimating the cell number on four subsequent days. The cells were detached by trypsin treatment and the cell number was measured in an automated cell counter. Means ± SEM of three independent experiments performed in triplicate are shown.

Reduced cell aggregation of EGFP/Lef-1 Δexon VI-expressing PANC-1 correlated with an elevated cell migration (Fig. 6b) and invasion through collagen-coated transwell chamber inserts (Fig. 6c). In contrast EGFP/Lef-1 FL-expressing cells exhibited migration and invasion rates comparable with mock transfected controls. Expression of the dominant negative Lef-1 Δexon VI ΔHMG did not accelerate cell migration or invasion through collagen. These results point to a Lef-1 Δexon VI-specific enhancement of pancreatic cancer cell migration.

Lef-1 FL increases the proliferation of pancreatic tumour cells

The effect of Lef-1 FL-induced c-myc and cyclin D1 expression on cell cycle progression was investigated in proliferation experiments. After transient expression of EGFP, EGFP/Lef-1 FL or EGFP/Lef-1 Δexon VI, equal numbers of cells were seeded and the number of cells was determined after different periods of time. PANC-1 or PaTU8902 cells expressing Lef-1 FL exhibited a 3.9-fold increase in cell number after 72 hours of incubation, whereas the number of Lef-1 Δexon VI- or EGFP-expressing cells increased only 2.2-fold and 2.1-fold, respectively (Fig. 6d). To verify these effects on cell proliferation, the number of mitotic figures was determined by DNA staining 36 hours after transfection with EGFP-tagged Lef-1 variants. To rule out different transfection efficiencies, only EGFP-positive cells were considered in this experiment. The results shown in figure 6D demonstrate that PaTU8902 cells expressing EGFP/Lef-1 FL exhibited a significantly higher number of dividing cells (13.6 mitotic figures per 100 cells, SD = 2.2) in contrast to EGFP/Lef-1 Δexon VI- or EGFP-expressing cells (6.3 mitotic figures per 100 cells, SD = 1.3 or 6.8 ± 1.6, respectively) (Supporting Information Fig. 2b). This increase of mitotic figures in EGFP/Lef-1 FL-expressing cells was statistically significant using Student's t test (p < 0.05).

Discussion

Lef/TCF transcription factors play multiple roles during ontogenesis as well as in certain types of cancer. In the present study, we have characterised a new Lef-1 isoform expressed in pancreatic ductal adenocarcinoma which lacks the exon VI compared to the full length transcript. Expression of Lef-1 Δexon VI was increased in advanced pancreatic tumour stages and in pancreatic adenocarcinoma compared to healthy tissue. This correlation was reflected by several pancreatic carcinoma cell lines such as AsPC-1, Capan-2 or PaTU8988t which express the Lef-1 Δexon VI isoform and possess a high metastatic capacity after injection into nude mice.34, 35 PANC-1 and BxPC-3 pancreatic carcinoma cells which show a low metastatic potential express only the full length Lef-1 FL isoform.36, 37

In the present work, we show that the Lef-1 Δexon VI isoform regulates the activity of specific Lef-1 target genes but differs in some aspects from full length Lef-1. Lef-1 Δexon VI inhibited the expression of E-cadherin, but stimulated the activity of the fibronectin promoter. By contrast, full length Lef-1 did not alter transcriptional activities of these target genes. These diverse effects were verified using different experimental setups like RT-PCR, Western blotting and luciferase reporter assays as demonstrated in figures 2, 3, 4 and 5. These data emphasise the complexity of Lef-1 signalling and its regulation of gene expression. We demonstrate that the smaller isoform Lef-1 Δexon VI predominantly regulates E-cadherin gene expression at least in pancreatic epithelial cells. Chromatin precipitation assays demonstrated that Lef-1 Δexon VI binds predominantly to or near the classical E-box elements of the E-cadherin promoter and contributes to the repression of E-cadherin transcription which is observed in some Lef-1-expressing epithelial cells. Although we can not exclude that the exact binding sequence of Lef-1 Δexon VI vary from that identified by Nawshad et al. (2007) or that other factors are necessary for the regulatory effect of Lef-1 Δexon VI. In agreement with data published by Nawshad and colleagues26 this effect of Lef-1 Δexon VI did not depend on binding to β-catenin, pointing to a β-catenin independent role of Lef-1 Δexon VI in the regulation of E-cadherin expression. Most interestingly, inhibition of TGFβ-signalling by a pharmacological inhibitor interfered with Lef-1 Δexon VI-induced repression of the E-cadherin promoter activity. These data support the hypothesis that a cross-signalling between the TGFβ-induced signalling and Lef-1 Δexon VI contributes to regulation of E-cadherin gene expression during carcinogenesis as suggested by Nawshad et al.26

The presence of different Lef-1 isoforms may explain the complex and heterogeneous contribution of this transcription factor family to tumour development.13 In addition to the Lef-1 FL isoform, an alternative mRNA transcript of Lef-1 has been isolated from colorectal cancer samples and was subsequently identified in other tumours as well.13, 38, 39 By using an alternative promoter and translation start this isoform lacks exons I and II resulting in the loss of the β-catenin binding site.6, 39 It acts as a transcriptional repressor predominantly expressed in normal colon epithelial cells.40 A third Lef-1 mRNA transcript encodes for a C-terminal truncated Lef-1 protein devoid of the HMG-box and the nuclear localisation signal which are both localised at the C-terminus.41 This human Lef-1 isoform of unknown function is the result of the use of an alternative exon VIII containing a premature stop codon.41

The splice variant of Lef-1 without exon VI, characterised in the present manuscript, was described before by Hovanes et al. as a putative variant6 but not further characterised. This isoform lacks parts of the context-dependent regulatory domain (CRD) and possesses different regulatory activities compared to full length Lef-1. Lef-1 Δexon VI inhibited cell-cell adhesion resulting in enhanced cellular migration, whereas the full length Lef-1 isoform stimulated cell cycle progression and enhanced cell proliferation at least in pancreatic carcinoma cells. The exon VI of Lef-1 contains 84 base pairs which may act as binding domain for or contributes to the modulation of binding affinity of diverse Lef-1 interaction partners. It might be speculated that the newly characterised isoform Lef-1 Δexon VI, with the slightly modified CRD domain, possess a different binding affinity to Lef-1-interacting coactivators or corepressors leading to the observed differences between the analysed Lef-1 isoforms. Lef-1 Δexon VI-specific cofactors may be more effective in binding to DNA sequences containing less highly conserved Lef-1-binding sequence, such as these in the E-cadherin promoter. This may contribute to fine-tuning of Lef-1-regulated cell functions such as cell proliferation or cell migration. One candidate as Lef-1 Δexon VI-binding protein might be Smad2/Smad4. Cooperation between both proteins/complexes has been suggested to be involved in the regulation of E-cadherin promoter activity by Nawshad and colleagues.26 Our findings that an inhibition of the TGFβ receptor kinase suppresses the Lef-1 Δexon VI-induced repression of E-cadherin transcriptional activity support this hypothesis.

In Xenopus laevis, a similar alteration of the Lef/Tcf-family member XTCF-4 which influences the transcriptional activity of XTCF-4 has been described by Gradl et al.42 The alternatively spliced motives LVPQ and SLVSS, flanking TCF-4 exon IVa, which corresponds to exon VI of human Lef-1,43 regulate the responsiveness of XTCF-4 target genes. The presence of these motives results in an inhibitory XTCF-4, called isoform A.42 An isoform without flanked exon IVa acts as an activator of targets such as fibronectin and siamois.42 These data demonstrate that alternatively spliced Lef/TCF transcription factors can act as repressors or activators of gene expression. The binding of different coregulators to the Lef-1 isoforms may explain the heterogenity of transcriptional control. Proteins which interact with members of the Lef/TCF family most often bind to these in the highly conserved regions, e.g., the HMG-box, as it is known for Smads, Cdx1 or PIAS.44, 45, 46 Other binding partners interacting specifically with only one member of the Lef/TCF-family interact preferentially with the CRD of Lef/TCFs. Our preliminary data suggest that regulation of c-myc by Lef-1 FL comprise the histone acetyltransferase p300 (Supporting Information Figure 2a). By using an oligonucleotide precipitation assay containing a sequence with the Lef-1-binding site of c-myc, p300 and β-catenin were coprecipitated with the c-myc promoter. These data support the hypothesis by Daniels and Weiss (2002) that p300 interacts with Lef-1 via β-catenin in a transcriptional regulation complex.47

Regarding the role of Lef/TCFs in the development of different types of cancer, the best analysed example is the role of TCF-4 in colon cancer.48, 49 Among the targets of TCF-4, c-myc and cyclin D1 seem to be essential for transformation of colon cells.8, 50 Lef-1 was also identified as a mediator in leukaemia as well as colon, breast and skin cancer.2, 51, 52 Our data support the role of Lef-1 in tumourigenesis, especially of pancreatic cancer, by regulating c-myc expression. Luciferase reporter assays confirmed that the c-myc promoter activity can be increased by full length Lef-1. Vogelstein and coworkers7 identified two Lef-1/TCF-binding sites in the c-myc promoter which are in suspicion to mediate the Lef-1 FL effects observed in this report.

Limited data are available about the importance of Lef/TCF transcription factors and Wnt signalling in pancreatic cancer.53 Recent data published by Ripka et al. suggest that Wnt5A-induced activation of the Lef/TCF-β-catenin-dependent pathway possesses tumour-promoting effects in pancreatic carcinoma cell lines and results in enhanced cell migration and invasiveness.54 Recently, we have demonstrated that alterations of the extracellular environment of pancreatic tumour cells, namely the increase of fibrilliar collagen, induces nuclear localisation of β-catenin and increased activity of the Lef/TCF-dependent reporter TopFLASH pointing to a role of Lef-1 in invasive growth of pancreatic carcinoma.55

Taken together, the present data show that an enhanced expression of Lef-1 in pancreatic cancer correlates with more advanced tumour stages. A new splice variant of Lef-1 lacking exon VI differentially regulates target gene expression compared to full length Lef-1. The presence of Lef-1 Δexon VI in pancreatic cancer cells increases cellular invasion by affecting E-cadherin expression, cell aggregation and migration. The existence of different Lef-1 isoforms as well as differential binding of cofactors by these isoforms may explain the diverse effects of Lef-1 in the contexts of initiation and progression of carcinoma and particularly pancreatic cancer.

Acknowledgements

We thank I. Baum, C. Laengle, C. Ruhland and K. Reutlinger for excellent technical assistance and Dr. K. Giehl for discussion and critically reading the manuscript. The work was supported by the Deutsche Forschungsgemeinschaft (SFB 518, B4) and by the MD program of the medical faculty of the University Ulm to SJ.