Introduction
Head and neck squamous cell carcinoma (HNSCC) is one of the most common types of human cancer (1). Only 40–50% of patients with HNSCC will survive for 5 years after diagnosis and treatment. Despite the advent and use of targeted drugs in the recent past, the survival has not markedly improved. Most patients develop locoregional recurrence, distant metastases or second primary tumors. The underlying cellular and molecular mechanisms that contribute to the initiation and progression of HNSCC have not been completely delineated (2). Recent advances in the understanding of the oncogenesis of HNSCC have revealed multiple molecular deregulations, with PTEN/PI3K/Akt/mTOR being the most frequently altered signaling pathways. Loss of PTEN expression (phosphatase and tensin homolog deleted on chromosome 10) was found in 29% of tongue cancers and loss of heterozygosity identified in 40% of HNSCC (3–5). A decrease in PTEN expression is sufficient to induce tumor progression (6,7). PTEN converts the phosphatidylinositol-3,4,5-trisphosphate (PIP3) in the cytoplasm to phosphatidylinositol-4,5-bisphosphate (PIP2), thereby directly antagonizing the activity of PI3K (8). Its inactivation is common in cancer and results in constitutive activation of the PI3K/AKT pathway. PTEN also inhibits SHC phosphorylation, reducing the activation of Ras/MAP-kinase (MAPK) pathway (9). Nevertheless, more attention has been focused on its capacity to antagonize the growth and proliferation-promoting actions of PI3K and AKT.
More recently, the protein phosphatase activity of PTEN has received increased scrutiny in aberrant cellular processes, particularly regarding cell motility, invasion and migration. Indeed, it has been demonstrated that PTEN modulates cell migration and invasion by negatively regulating the signals generated at the focal adhesions, through direct dephosphorylation and inhibition of FAK (10,11). Furthermore, PTEN can modulate integrin-mediated focal adhesion formation and cytoskeletal reorganization (12).
The effects of loss of PTEN have primarily been evaluated in cancer cell lines that have other oncogenic mutations and have lost their epithelial phenotype (13). Such an approach has made it difficult to determine the phenotypes directly conferred by the loss of PTEN and to define the stages of tumorigenesis that are specifically influenced by the PTEN loss. To further characterize the role of PTEN in HNSCC carcinogenesis, we have selected two cell clones expressing a small hairpin RNA targeting a specific sequence of PTEN previously validated (14) and the invasive ability of these two cell clones was evaluated both in vitro and in vivo. Our results showed that the tumors originating from PTEN deficient cells have a molecular signature and a phenotype suggesting an undergoing epithelial to mesenchymal transition (EMT). To our knowledge, our study is the first to investigate a role of PTEN in the EMT in HNSCC.
Materials and methods
Cell culture and transfections
CAL 27 cells, a human head and neck squamous cell carcinoma (HNSCC) cell line, were used (CRL-2095™, LGC Standards, Molsheim, France). The cell line was cultured in phenol-red free RPMI-1640 (Invitrogen, Saint Aubin, France) supplemented with 10% of heat-inactivated fetal calf serum (Sigma-Aldrich, Saint-Quentin Fallavier, France) at 37°C in a humidified atmosphere of 5% CO2 in air. pSIREN-RetroQ expression vectors containing shPTEN and a pSIREN-RetroQ scramble shRNA vector were transfected in CAL 27 using Jet-PEI (Polyplus Transfection, Illkirch, France) according to the manufacturer’s protocol. Stable, pooled populations were then generated under puromycin selection (0.5 μg/ml; Sigma-Aldrich) and clones were isolated by limiting dilution. Two transfected CAL 27 cell clones named shPTEN1 and shPTEN2 were used in this study and compared to scramble vector-transfected cells (shCTL) or parental CAL 27 cells.
Cell proliferation
The effect of PTEN inhibition on cell proliferation was evaluated by trypan blue exclusion assay. Briefly, cells were cultured in 6-well dishes (2×104 cells/well) over 10 days. After the set time, cells were detached using a 0.25% trypsin-EDTA solution (Sigma) for 7 min at 37°C and resuspended in complete medium. Cells were then stained with 0.4% (w/v) trypan blue solution and viable cells were counted twice using a Malassez hemocytometer.
Cell adhesion and invasion
For cell adhesion, cells were stained with Hoechst 3342 for 30 min at 37°C before trypsination. Then cells were plated at a density of 2×105 cells/wells on fibronectin, collagen I, laminin and vitronectin coated wells (at a concentration of 10, 20, 4 and 0.2 μg/well, respectively). Cells were allowed to attach for 0.5 and 4 h. Then, the medium containing non-adherent cells was gently removed. Cell adhesion was measured by fluorescence intensity with a microplate reader at 460 nm (Twinkle LB 970, Berthold Technologies, Thoiry, France).
Cell invasion assay was performed in modified Boyden Chambers with 8-μm pore filter inserts (BD Falcon, Erembodegem, Belgium) coated with Matrigel™ (BD Falcon). Cell lines were added at 2×105 cells per well to the upper chambers in 300 μl of serum free media. The lower chambers were filled with 500 μl of medium containing 10% (v/v) fetal bovine serum. After 24-h incubation at 37°C, non-invading cells were removed from the upper chambers with a cotton-tipped swab. Invading cells in the lower chambers were fixed, stained with 0.5% (w/v) crystal violet solution and then lysed using 10% (w/v) SDS solution. The absorbance at 560 nm was measured in a microplate reader (Multiskan Ascent, Thermo Electron Corporation, Saint-Herblain, France).
Analysis of proteins
For analysis of PTEN, E-cadherin, vimentin and integrin-β1 expression, western blot analysis was realized as previously described (14). The following primary antibodies against human PTEN (M3627, DakoCytomation, Glostrup, Denmark, 1:1,000 dilution), E-cadherin (#3195, Cell Signaling Technology, Danvers, MA, USA, 1:1,000 dilution), vimentin (#3390, Cell signaling Technology, 1:1,000 dilution), integrin-β1 (#4706, Cell Signaling Technology, 1:1,000 dilution) and β-tubulin (sc-9104; Santa Cruz Biotechnology, Santa Cruz, CA, USA, 1:1,000 dilution) were used. Quantification of relative band densities was performed using densitometer (LAS4000 Imager FujiFilm, Tokyo, Japan) and β-tubulin was used as the internal control.
For analysis of phosphoproteins regulated by PTEN, Bioplex phosphoprotein array based on multiplex sandwich bead immunoassay was used as previously described (14). Fluorescent capturing beads coupled to antibodies directed against phosphorylated-AKT and phosphorylated-ERK 1/2 (Bioplex phosphoprotein assays, Bio-Rad, Marnes-la-Coquette, France) were added into each well containing 6 μg of protein.
Screening of soluble proteins secreted by the cells after 48 h of culture was realized using the proteome profiler array assay (R&D Systems, Lille, France) as previously described (15). Factors involved in cell invasion (pro-invasive factors, MMP9, uPA; and anti-invasive factors, TIMP1, PAI-1) were detected in significant concentrations.
Tumor growth
All experiments were performed in accordance with national animal care guidelines (European Commission directive 86/609/CEE; French decree no. 87–848). Eight-week-old female athymic NMRI-nu (nu/nu) mice (Janvier Laboratories, Saint Berthevin, France) were randomly divided into 4 groups (n=12). All groups received subcutaneous injections of Matrigel and glucose 0.5% (1:3) containing 1×107 cells. Tumor growth was evaluated, twice a week, by measuring the length and width of the tumor mass with calipers. Tumor volumes were calculated by the modified ellipsoidal formula: V = 1/2 (length × width2).
Histology and immunohistochemistry
Tumor fragments from xenografts harvested 60 days after cell injection were fixed in 4% (w/v) formaldehyde, pH 7.4 for 16 h and embedded in paraffin. Five micrometer sections were used for hematoxylin and eosin and immunochemical staining, as previously described (16). Primary antibodies, sources and dilutions, as well as the antigen retrieval solutions and detection systems of primary antibodies are listed in Table I. Each assay included a negative control consisting of the staining with secondary antibody but in absence of a primary antibody.
Statistical analysis
All results were given as mean ± standard error of the mean (SEM). Non-parametric Mann-Whitney U test was employed to determine the statistical significance with a limit set to p<0.05 using GraphPad Prism 5 (GraphPad Software, La Jolla, CA, USA).
Results
Decrease of PTEN expression activates AKT signaling pathway and promotes cell survival but not cell growth in vitro
We first evaluated the expression of the stable inhibition of PTEN in CAL 27 cell lines (Fig. 1A). We assessed the phosphoproteins expression directly regulated by the phosphatase activity of PTEN, such as pAKT and pERK 1/2. The level of expression of pAKT was decreased by 48% in the clone shPTEN1, and 38% in the clone shPTEN2 (Fig. 1B), while the expression of pERK 1/2 was similar to the control cell line (Fig. 1C).
As the semi-quantitative analysis of phospho-proteins was measuring a 3-fold increase in AKT activation in the PTEN deficient CAL 27 cells, cell growth and survival were assessed by a proliferation assay (Fig. 2). All cell lines were in an exponential phase from 48 to 120 h. The doubling time (calculated on these time points) of shPTEN1 and shPTEN2 was close to those of shCTL and CAL 27 (Table II). Nevertheless a slope change occured at 120 h (Fig. 2, day 5), when shPTEN1 cells engaged a plateau-like phase, while shPTEN2 cells kept proliferating (day 8) without showing the plateau phase. After this time-point, the number of shCTL and CAL 27 cells was half the number of shPTEN1 and −2 cells (day 10, p<0.001 compared to the shCTL cell line) indicating that cells not only stop proliferating, but also could die. Contrarily, shPTEN cells that were less sensitive to contact inhibition survived and further proliferated. This result is consistent with the overactivation of pAKT, promoting cell survival.
Decrease of PTEN expression is associated with loss of cell adhesion in vitro
In culture, the morphology of PTEN deficient cells suggested a loss of the epithelial phenotype and of cell adhesion. The attachment assay performed at 0.5 h after plating (Fig. 3A) indicated an interaction of the PTEN-deficient cells significantly increased with specific extracellular matrix (ECM) proteins, notably shPTEN1 cells with fibronectin. This interaction is only transient, as at 4 h, shPTEN cells are not adhering as much compared to the parental and shCTL cell lines (Fig. 3B), notably on laminin and vitronectin. The results suggest that the loss of adhesion was dependent on the inhibition of PTEN, with a loss of adhesion on fibronectin of 18% in shPTEN1 cells and 32% in shPTEN2, when compared to the shCTL cell line. Associated to the loss of cell adhesion, the expression of β1-integrin was found to be upregulated by 25 and 30%, respectively, for shPTEN1 and shPTEN2, as compared to shCTL cells.
Decrease of PTEN expression stimulates cell invasion in vitro
Concomitantly to a loss of adhesion, the increase in β1-integrin expression in shPTEN cell lines indicated a gain in mobility. However, a migration assay such as wound healing could not be performed, because shPTEN cells would reach confluency without covering the surface evenly, thus preventing an accurate assessment of wound closure. However, the invasiveness of the cell lines was evaluated using a modified Boyden chamber assay. This assay established that shPTEN1 cells were significantly more invasive than the other cell lines (Fig. 4A), contrary to shPTEN2. In line with this result, the proteolytic proteins MMP9 and uPA were found in higher concentrations in the media of shPTEN1 cells compared to shCTL cells (Fig. 4B). Moreover, TIMP1, inhibitor of MMP9, was not upregulated for shPTEN cells contrary to PAI-1, inhibitor of urokinase-type plasminogen activator (uPA).
As markers of the epithelial to mesenchymal transition, the expression of E-cadherin and vimentin was evaluated in vitro. As expected in epithelial cells, vimentin was not detected, in any of the CAL 27 cell lines. However, E-cadherin expression was significantly different in the PTEN deficient cells, reduced by 2 to 3 times compared to the shCTL cells (Fig. 4C).
Decrease of PTEN expression impairs tumorigenicity of HNSCC cells in vivo
Injected subcutaneously in nude mice, the parental and shCTL CAL 27 cell lines established tumors within 3 weeks, the tumors of shPTEN cells were palpable but the volume remained smaller than 100 mm3 for 4 weeks (Fig. 5A). Even though there was a significant difference in the growth rate between the shCTL cells and the parental CAL 27 cells, the tumorigenicity of PTEN deficient cells in vivo was significantly reduced. The staining of Ki67 in these tumors confirms that the slow growth is a consequence of a lower proliferation of shPTEN cells (Fig. 5B).
Decrease of PTEN expression induces the formation of islets of well-differentiated tumor cells in vivo
Xenografts of shPTEN cells grew as well-differentiated squamous cell carcinomas with readily apparent keratinization as observed by the cytokeratin staining (arrows on Fig. 6). Some interesting differences can be found in the pattern of staining for PTEN expression. The localization of PTEN in the tumors originating from shCTL cells was predominantly cytoplasmic in epithelial cells. In contrast, the epithelial cells of shPTEN tumors (which still express 40 to 50% of PTEN protein compared to the shCTL cells) were weakly stained for PTEN, and interestingly the localization was also nuclear for shPTEN1 cells (Fig. 6).
Decrease of PTEN expression is associated with an epithelialmesenchymal transition phenotype in vivo
The expression of molecular markers of EMT: E-cadherin and vimentin, were also evaluated on sections of formalin-fixed tumors (Fig. 7). The membrane E-cadherin staining detected in shCTL and shPTEN1 tumors is typical of an epithelial phenotype (Fig. 7). Interestingly, the staining in shPTEN2 tumors is more diffused. As expected, vimentin was detected in the cytoplasm of the least differentiated cells located at the periphery of the main tumor sheets and cell islets. However, in shPTEN1 tumors, vimentin is stained in the cytoplasm of isolated tumor cells (Fig. 7). This is consistent with the presence of less differentiated territories in these tumors, as well as the presence of numerous isolated tumor cells observed in the connective tissue at a distance of the main tumor masses.
Discussion
A crucial step in cancer progression is the transforming event turning cells into an overt carcinoma with invasive and metastasic properties. Acquisition of such properties occurs in a multistep and complex process, as cells have to lose adhesion and escape the tumoral environment to disseminate and establish secondary tumors. Epithelial to mesenchymal transition is a modality that can be used by tumors to progress towards dissemination and metastasis, but is still challenging to study in in vivo models or through patient tissue banks (17).
Whereas 50% of HNSCC patients will suffer from recurrent tumor, new primary tumors or metastasis within 5 years after diagnosis (even after treatment), the mechanisms behind progression have not been elucidated. In a recent review, Leemans et al (18) have described invasion and early metastatic dissemination of HNSCC as a ‘biologically-driven’ process, highlighting the absence of genes known to induce it. Although the authors review how epithelial to mesenchymal transition could play a role in HNSCC dissemination, notably through the possible implication for the neurotrophic factor receptor NTRK2 and its ligand the brain-derived neurotrophic factor (BDNF), they do not provide any further evidence to substantiate an actual EMT process in HNSCC. This emphasizes the lack of experimental data documenting the process of EMT in HNSCC progression. Our study relates, for the first time, to the common loss of PTEN expression in the epithelial-mesenchymal transition in HNSCC. In our model, the decrease of PTEN expression promoted cell survival in vitro, most likely through an induced activation of AKT, as previously described (14). In addition, the PTEN deficient cells showed a loss of adhesion concomitant to an increased expression of β1-integrin, which indicated a gain in migrating properties. Our results also demonstrate that PTEN deficient cells were more invasive than the parental cells, associated to levels of MMP9 and uPA significantly upregulated in shPTEN1 cells. Accordingly, the lower concentration of protease expression detected in shPTEN2 media could explain the discrepancy with shPTEN1 cells in invasiveness. Indeed, uPA-mediated proteolysis was shown to promote tumor invasion, while PAI-1 has been described to compete with integrin binding to vitronectin (19), which is consistent with the data shown here in the adhesion assay.
Together, our results show that the PTEN deficient cells had a survival advantage and were more invasive. Morphological changes suggested the acquisition of a mesenchymal phenotype, which was confirmed by a decrease in the expression of E-cadherin. Similarly, the morphological observations on sections of formalin-fixed tumors gave some evidence of a diminution of the membrane expression of E-cadherin in PTEN deficient tumors, which also showed the presence of islets of well-differentiated cells. PTEN appears to be essential in maintaining the balance between cell growth, differentiation and malignancy of HNSCC. It has been previously demonstrated that PTEN nuclear localization occurred preferentially in differentiated cells or resting cells (20). Nuclear localization of PTEN may also have distinctive effects on cell growth, apoptosis and differentiation (21,22). Very interestingly, we show that the remaining PTEN protein in tumors originating from shPTEN cells was preferentially localized in the nuclei.
To summarize, our data provide evidence of an epithelial-mesenchymal transition occurring in HNSCC PTEN deficient cells. However, the in vivo model for this study was established with the cells already repressed for PTEN expression, which significantly compromised tumorigenicity. The staining of Ki67 as well as in vitro cell growth showed that it was not due to the proliferation, but could rather be caused by loss of cell adhesion and invasiveness as well as cell differentiation. According to the invasive properties of the PTEN deficient cells, the tumors formed were ulcerative. Interestingly, proliferation in our model, confirmed in vivo, is very similar to the heterozygous loss of PTEN induced in a model of breast cancer cells developed by Vitolo et al (23). The authors show a lower proliferation and a resistance to anoikis (programmed cell death after to cell-detachment of the extracellular matrix).
It would be of interest to evaluate EMT in a conditional knockdown model for PTEN in HNSCC, especially with the loss or reduction of PTEN expression being a late molecular event in the carcinogenesis of HNSCC (24). Similarly, the EMT being a very dynamic process, further live imaging experiments would greatly contribute to elucidating the exact role of PTEN in EMT in this type of cancer.
A link between the expression of PTEN and EMT has been described previously (25), notably through the interaction of PTEN with the β-catenin/wnt pathway and the downregulation of E-cadherin (26–29). Furthermore, it has been shown that PTEN is involved in stabilization of E-cadherin-mediated adherent junctions (30) and cell polarity (31). We are confident that, with further investigations, PTEN could be proposed as a candidate marker for early dissemination in HNSCC, and possibly the risk of recurrence. Our data are in line with clinical reports (32,33). The most recent one is a genotyping approach conducted on 440 samples from HNSCC patients, in which a genetic variant has been correlated to a high risk of recurrence (33). Similarly, Lee et al (5) already established a link between loss of PTEN expression and advanced disease. Interestingly, in their study, 75% of the patients with lack of PTEN expression had recurrence, metastasis or death (vs. 41% in the PTEN positive group).
Our study provides evidence for a new description of role of PTEN in the progression of HNSCC tumors through the EMT. We propose PTEN as a candidate biomarker to predict high risk of recurrence and metastasis in HNSCC. The nuclear localization of PTEN should also be considered in the future histological analysis and may be related to the EMT phenotype.