Background
The Wnt/β-catenin pathway plays important roles in morphogenesis, normal physiological functions, and tumor formation. At the molecular level, β-catenin is involved in two apparently independent processes, cell-cell adhesion and signal transduction [
1]. In the absence of a mitotic signal, β-catenin is sequestered in a "destruction complex" which consists of the adenomatous polyposis coli (APC) gene product, casein kinase 1 (CK1), a serine threonine glycogen synthetase kinase (GSK-3β), and axin, an adapter protein [
2]. This destruction complex is phosphorylated and degraded by the ubiquitin-proteasome system [
2]. β-catenin also plays a role in the transcription activation pathway [
3,
4]. Following stimulation of mitosis signal, β-catenin accumulates in the cytoplasm, moves to the nucleus, and then binds to a member of the TCF/LEF-1 family of transcription factors that modulate expression of TCF/LEF-1 target genes [
5‐
7]. Previously, we and others reported that aberrant expression of β-catenin was common in oral cancer and this change correlated with the malignancy index and patient prognosis [
8,
9]. However, the molecular mechanisms that lead to aberrant expression of β-catenin in oral cancer are unclear, and the mechanisms by which β-catenin promotes activation of target genes are also not well understood.
Certain mutations of APC or β-catenin increase β-catenin signaling, leading to overexpression of oncogenes and promotion of neoplastic growth [
10‐
15]. However, for some cancers, β-catenin accumulates in the nucleus even though mutation of β-catenin or APC is rare. For example, in endometrial cancers, 12 of 20 cases (60%) exhibited β-catenin accumulation in the nucleus, but only two of these cases had mutations in the β-catenin gene [
16]. In hepatocellular carcinomas, nearly 50% of cases exhibited nuclear accumulation of β-catenin, but APC mutation was very rare and only 16-26% of cases had mutations in β-catenin [
10,
17‐
19]. Similar findings have been reported for oral cancer [
8]. Therefore, it is possible that mechanisms other than mutation are involved in the aberrant β-catenin expression observed in tumors.
Recent reports have suggested that receptor tyrosine kinases (RTKs) can regulate β-catenin function [
20,
21]. Epidermal growth factor receptor (EGFR) is a member of the receptor tyrosine kinase family, and overexpression of EGFR is associated with poor prognosis and progression of many human cancers, including oral cancer [
22,
23]. At the molecular level, stimulation of EGFR induces intrinsic tyrosine kinase activity and cellular signaling that results in cell growth and proliferation. EGFR stimulation is associated with perturbation of E-cadherin-mediated cell adhesion, development of fibroblast-like morphology, and increased cell motility in certain tumors [
24‐
26]. Moreover, EGFR interacts with the β-catenin core region and induces tyrosine phosphorylation of catenins in several types of tumors [
27,
28]. This raises the possibility that EGFR signaling may play a role in the regulation of β-catenin. It is not yet known whether EGFR plays a role in the aberrant expression of β-catenin that is seen in oral cancer.
In the present paper, we describe the effect of EGFR signaling on the nuclear accumulation of β-catenin in oral cancer. This extends our previous research into the mechanisms that underlie aberrant accumulation of β-catenin.
Methods
Cell culture and reagents
All cell lines were maintained in DMEM or RPMI1640 media that were supplemented with 10% bovine serum and 1% gentamycin. Cells were maintained in a humidified atmosphere containing 5% CO2 at 37°C and the medium was changed three times per week. Cell lines were grown until they were 89-90% confluent. All cultures were negative for mycoplasma infection.
Recombinant human EGF was obtained from R&D Systems (Minneapolis, MN, USA), EGFR inhibitor (AG1478) from A.G. Scientific (San Diego, CA, USA), lithium chloride from Acros Organics Co. (Geel, Belgium), Erbitux from Merck (Darmstadt, Germany), mouse anti-E-cadherin and mouse anti-β-catenin from BD Transduction Lab (Lexington, KY, USA), phospho-GSK-3β (Ser-9), phorpho-β-catenin (Ser33/37/Thr41), EGFR antibody, and phosphor-tyrosine antibody from Cell Signaling Technologies (Beverly, MA, USA), anti-HDAC1, anti-cyclin D1, goat anti-rabbit IgG-HRP, donkey anti-goat IgG-HRP, and protein A/G Plus-Agarose immunoprecipitation reagent from Santa Cruz Biotechnology (CA, USA), anti-Suv39h1 (05-615), anti-acetyl histone H4 (06-88-66), anti-trimethyl-histone H3K9 (07-442), and anti-trimethyl-histone H3K4 (07-473) from Upstate Chemicon (Temecula, CA, USA), rabbit anti-mouse IgG conjugated to HRP antibody from Novus Biologicals (Littleton, CO, USA), anti-human EGFR and anti-CBP, anti-Lamin B1, and anti-alpha-tubulin from Abcam (Cambridge, UK).
Patients and tissue specimens
All specimens were obtained from the archives of Tri-Service General Hospital (Taipei, Taiwan) and included 112 samples of oral squamous cell carcinoma (HNSCC). The study design was approved by the Internal Review Board of Tri-Service General Hospital (TSGHIRB 095-05-116). More detailed information about the specimens was provided previously [
29]. A series of 5-μm sections were cut from each tissue block. A 5 μm flanking section was stained with hematoxylin and eosin (H&E) for pathological evaluation and to identify the cancerous and normal regions. Serial sections were used for immunohistochemistry (IHC).
Cell fractionation and Western blotting
Cellular fractionation was performed as described previously [
30]. Briefly, cells were washed twice with ice-cold phosphate-buffered saline, harvested by scraping with a rubber policeman, and lysed in a buffer (20 mM HEPES, pH 7.0, 10 mM KCl, 2 mM MgCl
2, 0.5% Nonidet P-40, 1 mM Na
3VO
4, 10 mM NaF, 1 mM phenylmethanesulfonyl fluoride, 2 μg/mL aprotinin). After incubation on ice for 10 min, the cells were homogenized by 20 strokes in a tightly fitting Dounce homogenizer. The homogenate was centrifuged at 1,500 ×
g for 5 min to sediment the nuclei. The supernatant was then centrifuged at 16,000 ×
g for 20 min, with the resulting supernatant considered the non-nuclear fraction. The nuclear pellet was washed three times with lysis buffer to remove contamination from cytoplasmic membranes. To extract nuclear proteins, isolated nuclei were resuspended in NETN buffer (150 mM NaCl, 1 mM EDTA, 20 mM Tris-Cl, pH 8.0, 0.5% Nonidet P-40, 1 mM Na
3VO
4, 10 mM NaF, 1 mM phenylmethanesulfonyl fluoride, and 2 μg/mL aprotinin), and the mixture was sonicated briefly to facilitate nuclear lysis. Nuclear lysates were collected after centrifugation (16,000 ×
g for 20 min at 4°C). Samples of each lysate were subjected to electrophoresis on an 8% SDS-polyacrylamide gel. Then, proteins were transferred to nitrocellulose membranes, immunoblotted with antibodies, and detected by electrochemiluminescence.
Preparation of cell lysates and immunoprecipitation
Cell monolayers were rinsed with 1× Tris-based saline (TBS) and then scraped into 1 mL of TBS. After a brief centrifugation, cells were solubilized in 1 mL of cell lysis buffer (150 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% TRITON® X-100 plus 1:100 protease inhibitor cocktail, P8340 from Sigma, and 1:100 phosphatase inhibitor cocktails, P5726 from Sigma). Before immunoprecipitation (IP), all samples were centrifuged at 12,000 × g for 30 minutes to remove insoluble cellular debris. For IP studies, lysates were pre-cleared for 1 h by use of protein A/G PLUS-agarose (sc-2003, Santa Cruz Biotechnology, CA, USA), incubated with antibodies at 4°C, and then treated with protein A/G PLUS-agarose for an additional 1 h. Immunoprecipitates were then washed 4 times with 1 mL TBS. After heating at 95°C for 10 minutes, proteins were resolved on SDS-PAGE, transferred to PVDF membranes for Western blot analysis, and immunoblotted with antibodies.
Luciferase reporter assays
SAS and OECM1 oral cancer cells were plated in 24-well dishes and incubated overnight at 37°C. The following day, cells were transfected with 1 μg of β-catenin-LEF/TCF-sensitive (TOP) or β-catenin-LEF/TCF-insensitive (FOP) reporter vector using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. On the following day, cells were washed with serum-free medium and treated with EGF (OECM1 cells; 100 ng/μL) or AG1478 (SAS cell; 20 μM). Reporter assays were performed using the luciferase reporter system (Promega, Madison, WI, USA).
Chromatin immunoprecipitation assay
The chromatin immunoprecipitation assay (ChIP) was performed using a kit from Upstate (Lake Placid, NY, USA) according to the manufacturer's instructions. Briefly, following treatment, cells were washed with PBS, cross-linked with 1% formaldehyde for 10 min, rinsed with ice-cold PBS, collected into PBS containing protease inhibitors, and then resuspended in lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris at pH 8.1 with 1% protease inhibitor cocktails). Cells were sonicated to produce 200-1000 bp of DNA fragments, followed by centrifugation to remove insoluble material. Samples were precleared for 1 h at 4°C with 60 μL of a 50% slurry of protein G agarose and salmon sperm. DNA immunoprecipitation was performed with indicated antibodies. Then, cross-links were reversed, and the bound DNA was purified by phenol:chloroform extraction. RT-PCR was performed using primers specific for human cyclin D1 promoter (5'-CCGACTGGTCAAGGTAGGAA-3' and 5'-CCAAGGGGGTAACCCTAAAA-3'). PCR reactions were run with PCR Master Mix (Promega), which consisted of 30 cycles of: 94°C × 30 s, 55°C × 30 s, and 72°C × 1 min, followed by 5 min at 72°C. PCR products were analyzed by 1.5% agarose gel electrophoresis, visualized with ethidium bromide, and then photographed. Images were saved as TIFF files and then analyzed with ImageJ
http://rsb.info.nih.gov/ij/. Signal intensities of the PCR data obtained from ChIP assays or from whole-cell lysates (Input DNA) were quantified from TIFF images by use of ImageJ, and then compared to the signal obtained for input control. Each ChIP experiment was repeated at least three times, and representative results are shown. Means and standard deviations (SDs) were calculated from the signal intensities.
Immunohistochemistry
Specimens that were embedded in paraffin blocks were cut into 5-μm sections. These were routinely stained with H&E for histological diagnosis, and additional sequential sections were selected for immunohistochemical studies. Immunodetection was performed with a standard DAKO EnVision stain system (Dako Corp, Carpinteria, CA, USA). Sections were dewaxed and subjected to antigen heat retrieval. Endogenous peroxidase activity and nonspecific binding were blocked by incubation with 3% hydrogen peroxide and nonimmune serum, respectively. Slides were then incubated sequentially with primary antibodies (16 h at 4°C) and DAKO labeled polymer secondary antibody (1 h at room temperature, then peroxidase-labeled polymer (30 min at room temperature). Diaminobenzidine hydrochloride (DAB) was used to visualize peroxidase activity. Then, sections were counterstained with hematoxylin and a cover slip was added prior to visualization.
Assessment of immunoreactivity
Using a semi-quantitative scale described previously [
8], the staining results of EGFR and cyclin D1 were classified as "high" or "low" staining. Briefly, in the clyclin D1 staining, tumors were evaluated as high if more than 10% of cells displayed nuclear staining and as low if otherwise. For the EGFR staining, scores representing the percentage of stained cancer cells were as follows: 0, no stained cells; 1, 1%-30%; 2, 31%-50%; and 3, >50%. Intensity was graded from 0 (no staining) to 3 (strong) in comparison with normal epithelium. Tumors were defined as high EGFR expression if the final score was 5 or 6 and as low if otherwise. The staining results for β-catenin were classified as membranous or cytoplasmic/nuclear, as in a previous report [
8]. Briefly, tumors were regarded as cytoplasmic/nuclear stain if unequivocal cytoplasmic and/or nuclear staining was present in at least one area of the tumor, and membranous stain if β-catenin was localized solely in the membrane. Immunostaining results were evaluated by two investigators (YSS and LCC) who had no prior knowledge of the histopathologic features of the tumor or the clinical status of the patient from whom the cell lines were obtained.
Statistical analysis
A χ2 test was used to assess the relationship of the results of the immunohistochemical determination of EGFR, β-catenin, and cyclin D1 expression and the clinical features of patients. P values less than 0.05 were considered statistically significant.
Discussion
Dysregulation of the Wnt/β-catenin signaling pathway has been linked to various human cancers, and this dysregulation is often associated with mutations in the β-catenin destruction complex components or in β-catenin itself [
26,
32]. However, β-catenin signaling is elevated in oral cancer cells even though mutations of APC and β-catenin are rare. This suggests that alternative mechanisms may contribute to β-catenin dysregulation. The present study demonstrated that the EGFR signal participates in the dysregulation of β-catenin in oral cancer. First, we found that the EGFR signal stabilized β-catenin and enhanced β-catenin nuclear accumulation by phosphorylated regulation. Moreover, we also showed that histone markers of open or repressed chromatin control the expression of cyclin D1, a β-catenin target gene. Finally, our study of oral cancer patients suggests that β-catenin-mediated cross-talk between EGFR and Wnt signaling may underlie the effect of EGFR during tumor development.
Numerous cell signals can impact β-catenin function. It was recently demonstrated that numerous oncogenic tyrosine kinases promote accumulation of β-catenin in the nuclei of different types of cancer [
33‐
36]. EGFR is the most commonly overexpressed receptor tyrosine kinase in oral cancer [
22]. The present study showed that an activated EGFR signal decreased membrane-bound β-catenin, increased nuclear accumulation of β-catenin, and induced mesenchymal cell morphology. This result was consistent with previous reports that the EGFR signal is associated with perturbation of E-cadherin-mediated cell adhesion, acquisition of fibroblast-like cell morphology, and increases in cell motility that are presumably related to tumor invasion and metastasis [
37,
38]. β-catenin plays a critical structural role in cadherin-based cell-cell adhesion and is also an essential coactivator of Wnt-mediated gene expression. The extent to which β-catenin participates in these two functions is controlled by the availability of β-catenin binding partners, and there is increasing evidence that these binding interactions are regulated by phosphorylation. For example, binding of β-catenin to E-cadherin and to α-catenin was substantially reduced when tyrosine in β-catenin was phosphorylated by EGFR [
27,
39]. Moreover full activation of GSK-3β generally requires phosphorylation of Tyr-216, whereas phosphorylation of Ser-9 inhibits GSK-3β activity. Therefore, our results suggest that the EGFR signal enhances accumulation of β-catenin in the nuclei of oral cancer cells directly, by phosphorylation of β-catenin, and indirectly, by stabilization of β-catenin through phorsphorylation and inhibition of GSK-3β.
The identification of many nuclear partners of β-catenin indicates that this protein functions as a transcription regulator by covalent modification of chromatin [
40,
41]. Many of these nuclear partners regulate chromatin structure by histone modification and chromatin remodeling. In the present study, the results of our ChIP assay demonstrated that an activated EGFR signal greatly increased the amount of CBP/P300 coactivator and reduced the amount of HDAC1 and Suv39h1 in the regulatory element of cyclin D1. A previous study showed that the central repeats of β-catenin (span R3-R10) is the region that interacts with TCF [
42]. In the absence of a nuclear β-catenin, TCFs recruit Groucho (TLF1 in mammals), a long-range chromatin repressor that functions with histone deacetylases (HDACs) to compress local chromatin and inhibit transcription [
43,
44]. Upon stimulation, β-catenin enters the nucleus and competes with Groucho for TCF binding, thus replacing the repressor with an activation scaffold [
45]. Our results showed that the extent of H3K4 methylation (H3K4me3) increased significantly following activation of cyclin D1 transcription by β-catenin in EGFR-activated cells, and that it gradually declined when the gene was inactivated in EGFR-inhibited cells. H3K4me3 is more common in active genes, and is believed to promote gene expression via recognition by transcription-activating effector molecules [
46]. A recent study showed that H3K4me3 also regulates another β-catenin target gene, c-myc [
47]. To the best of our knowledge, this is the first report to demonstrate β-catenin regulated cyclin D1 via histone modification/chromatin remodeling. Taken together, our results suggest that the EGFR signal promotes nuclear accumulation of β-catenin, which ultimately forms β-catenin-TCF complexes with histone-acetylating activity, and that these displace the repressor complexes. These β-catenin complexes remodel the chromatin structure of target gene promoters so that they are more accessible to the basal transcription machinery, thus enhancing transactivation of genes that leads to cellular responses.
The results of our experiments with cancer tissues corroborated the results of our experiments with cultured cells. In cancer tissues, EGFR expression correlated with the presence of nuclear β-catenin, and nuclear β-catenin correlated with the tumor malignancy index. This implicates the EGFR signal in mediating entry of β-catenin into the nucleus and progression of oral cancer. Our results are consistent with other studies which reported that nuclear β-catenin was present in 19-23% of oral cancer cells and associated with proliferation, invasiveness, and poor outcome of oral cancer [
8,
48]. In contrast, Gasparoni et al. reported that nuclear β-catenin was rare in oral cancer and found no clear association between intranuclear β-catenin and histopathological and malignancy indexes in vivo [
49]. The discrepancies between these studies could be explained by their use of different antibodies and methodologies. Although we did not find a close association between expression of nuclear β-catenin and cyclin D1, we did observe an association of nuclear β-catenin with the amount of cyclin D1 expression in some samples. This may be because multiple mechanisms regulate cyclin D1 expression in oral cancer cells [
50,
51]. For example, it is known that cyclin D1 amplification participates in overexpression of this gene in oral cancer [
52,
53]. Thus, in oral cancer, overexpression of cyclin D1 is more common than nuclear β-catenin expression (42% vs. 23%) [
8]. Taken together, EGFR activation is an alternative mechanism that induces β-catenin translocation to the nucleus of certain oral cancer cells. We suggest that measurement of the activation of this pathway may be a useful marker for measuring the progression of oral cancer.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
CHL performed the experiments and prepared the draft version of the manuscript. HWH and PHH participated part of the experiments and data analysis. YSS designed the experiments, supervised the project, and prepared the manuscript. All authors have read and approved the final version of the manuscript.