Fascin-1, Ezrin and Paxillin Contribute to the Malignant Progression and Are Predictors of Clinical Prognosis in Laryngeal Squamous Cell Carcinoma

Aims Fascin-1, ezrin and paxillin, cytoskeleton-associated proteins, have been implicated in several human cancers, but their role in laryngeal squamous cell carcinoma (LSCC) is unknown. We investigated the association of their expression and clinicopathologic factors and their prognostic value in LSCC. Materials and Methods Quantitative RT-PCR and western blot analyses were used to examine mRNA and protein levels in 10 fresh LSCC specimens and 10 corresponding adjacent normal margin (ANM) tissues from patients undergoing surgery in 2012. We used immunohistochemistry to retrospectively study 216 paraffin blocks of LSCC samples from patients (193 men) who had undergone surgery between 2000 and 2006 and had not received special treatment before the diagnosis. Univariate analysis of patient survival involved the Kaplan–Meier method. Multivariate analyses involved the Cox proportional hazards model. Results The relative mRNA and protein levels of fascin-1, ezrin and paxillin were significantly greater in LSCC than ANM tissue (P<0.05). The high expression of fascin-1, ezrin or paxillin was positively correlated with poor tumor differentiation, cervical lymph node metastasis (N+), and advanced clinical stage (III+IV) (P<0.05) but not sex or metastasis. In addition, a high expression of fascin-1 (P = 0.007) or ezrin (P = 0.047) was associated with advanced tumor stage (T3+T4). The expression of fascin-1 was higher in smokers than non-smokers (P = 0.019). A high expression of fascin-1, ezrin or paxillin was associated with poor prognosis. Conclusions Fascin-1, ezrin and paxillin may be prognostic of poor outcome with LSCC after surgery. Our study may lead to establishing new molecular therapeutic targets and/or prognostic biomarkers in LSCC.


Introduction
Head and neck squamous cell carcinoma, the 11th most common cancer among men worldwide [1,2], is one of the 6 tumors with a high incidence in China [3]. Laryngeal squamous cell carcinoma (LSCC), which originates from the laryngeal epithelium, has the second highest incidence of all head and neck squamous cell carcinomas, especially in the northern area of China, including Shanxi Province [4]. LSCC is a highly invasive and metastatic cancer [5]; this malignant behavior is one of the major reasons for treatment failure, which explains why the survival of LSCC patients has not improved much over years. Identifying some molecular indicators of the malignant behavior can be helpful for early prevention, diagnosis and treatment.
The incremental motility of malignant cells is a critical step in their migration, invasion and metastasis, which is regulated by reorganization of actin cytoskeleton and regulation of focal adhesion [6,7,8,9]. Fascin-1, ezrin and paxillin are essential components of these cellular structures.
Fascin-1 was originally described in the extracts of unfertilized sea urchin eggs as an actin-binding protein of 55 kDa. It is encoded by a gene located on chromosome 7p22 in humans [10]. The factor is responsible for actin bundle rearrangement, promoting tumor cell invasion and metastasis by increasing cell membrane protrusions, changing cell-cell and cell-extracellular matrix adhesion and regulating signal transduction pathways [10,11,12]. Fascin-1 overexpression was found associated with unfavorable prognosis in many malignant tumors such as colon cancer [11], gastric cancer [13], oral squamous cell carcinoma [14],esophageal squamous cell carcinoma (ESCC) [15], non-smallcell lung cancer (NSCLC) [16], and breast cancer [17]. Overexpression of fascin-1 was found associated with aggressiveness of LSCC but in a limited number of subjects [18,19].
Paxillin, a 68-kDa focal adhesion-associated protein, plays an important role in controlling cell spread and migration. Previous research has demonstrated paxillin at the interface between the plasma membrane and the actin cytoskeleton. It provides a platform for the integration and processing of adhesion and growth factor-related signals that are mainly involved in tumor metastasis and cell proliferation [32,33]. Paxillin is overexpressed in many carcinomas, including bladder cancer [33], ESCC [34,35], hepatocellular carcinoma (HCC) [36], NSCLC [37], salivary adenoid cystic carcinoma [38], and floor of mouth carcinoma [39].
Thus, fascin-1 is an actin-binding protein that promotes tumor cell invasion and metastasis, ezrin is a cross-linker of the actin cytoskeleton and the plasma membrane and plays a role in growth-factor-receptor signaling, and paxillin is a focal-adhesionassociated protein that plays an important role in controlling cell spread and migration. Thus, these proteins may contribute to the malignant behavior of LSCC. We aimed to detect the expression of these cytoskeleton-associated proteins in human LSCC and to correlate their expression with clinicopathological features and clinical outcomes.

Ethics Statement
This study was conducted in accordance with the Helsinki declaration. All patients signed a written, informed consent before surgery, acknowledging that they understood their rights and obligations. The study was approved by the Research Ethics Committee at Shanxi Medical University.

Clinical Samples and Patient Population
Ten LSCC tissues and 10 corresponding adjacent normal margin (ANM) tissues were obtained from patients undergoing surgery at the Department of Otolaryngology Head and Neck Surgery of the First Hospital of Shanxi Medical University in 2012. ANM tissues were isolated from surgical specimens at about 1 to 3 cm from the neoplastic edge. The fresh specimens were divided into 3 parts: 2 parts were frozen by use of liquid nitrogen for quantitative RT-PCR (qRT-PCR) and western blot analyses, and the other part was embedded in paraffin for hematoxylin and eosin staining to ensure the diagnosis of LSCC and ANM.
We examined data for 216 eligible patients (193 males) with LSCC who had detailed clinical records and had not received special treatment before the diagnosis. The patients underwent surgery in our department from 2000 to 2006. Two staff anatomical pathologists gave the diagnosis of laryngeal cancer. Patients did not receive radiotherapy or chemotherapy. Date of laryngeal cancer surgery, date of last follow-up and status at last follow-up (living, lost to follow-up or deceased) were recorded.
Tumor and clinical staging of paraffin-embedded tissues involved the tumor, node, metastasis (TNM) staging system of the Union for International Cancer Control (2010). The histological types of LSCC were determined according to the system of the World Health Organization.

Purification of Total RNA and qRT-PCR
Total RNA was purified from the frozen cancer tissues by use of a RNAiso Plus reagent kit (TaKaRa); RNA contents were determined 3 times in BioPhotometer plus (Eppendorf) to ensure OD 260 /OD 280 $1.8 and in 1% agarose gel electrophoresis to ensure integrity. Reverse transcription of total RNA (1 mg) involved use of the Rever Tra Ace qPCR RT Kit (TOYOBO); cDNA was amplified by qPCR with the SYBRH Green Realtime PCR Master Mix (TOYOBO) and the ABI PRISMH 7500 Sequence Detection System (Applied Biosystems). The cycling conditions included a 5-min initial denaturation step at 95uC, then 40 cycles at 94uC for 15 s, 60uC for 15 s, and 72uC for 32 s. The

Western Blot Analysis
Tissue protein lysates were obtained by use of the Tissue Protein Extraction Reagent and Proteasome Inhibition Mixture (Cwbiotech). Briefly, fresh tissue was rinsed with 4uC phosphate buffered saline (PBS) to remove blood on the surface, placed in liquid nitrogen for 2 to 5 s, then immediately ground into a fine powder. The powder was collected into an EP tube (Axygene) with cold lysis buffer and incubated on ice for 60 min. The lysate was centrifugated at 100006g, 4uC for 20 min in the Centrifuge 5702R (Eppendorf). Protein concentrations were determined by use of the bicinchoninic acid protein assay kit (Cwbiotech). Protein with 26 loading buffer (0.25 mol/L Tris-Cl, pH 6.8, 10% SDS, 0.5% bromophenyl blue, and 50% glycerol) was boiled for 5 min, loaded into 10% Tris-HCl polyacrylamide gels (80v, 50 min), then electrophoretically transferred to Immobilon-P Transfer Membrane (Millipore Corp.) and blocked with 5% nonfat milk at 4uC overnight and then incubated with primary antibodies for fascin-1 (mMAb IM20, 1:550, Vector Laboratories), ezrin (mMAb 3C12, 1:500, Abcam) and paxillin (mMAb 5H11, 1:700, Thermo Fisher Scientific) at 4uC overnight, then with rabbit anti-mouse IgG H+L (horseradish peroxidase conjugate [HRP], 1:4000) (Southern Biotech) for 2 hr at room temperature. Immunoreactive bands were detected and developed with use of Immobilon Western chemilum HRP substrate (Millipore Corp.) and X-ray film (Kodak). HRP-conjugated monoclonal mouse glyceraldehyde-3phosphate dehydrogease (GAPDH) from KangChen Biotech was used as loading control. The same analyses were performed 3 times. Analysis involved use of Quantity One v4.0 (Bio-Rad).

Antibodies and Immunohistochemistry Staining
Paraffin sections were dewaxed and re-hydrated in ethanol in descending concentrations (100%, 90%, 80%, 70%). For antigen retrieval, samples were processed in an autoclave as follows: fascin-1 for 2915 with sodium citrate pH 6.0; ezrin, for 290099 with sodium citrate pH 6.0, and paxillin for 291099 with EDTA, pH 9.0. Endogenous peroxidase activity was blocked by immersing tissue sections in 3% H 2 O 2 in methanol (v/v) at room temperature for 10 min and then washing with PBS. Nonspecific background staining was reduced by incubating sections with normal nonimmune goat serum (Boster Co.) for 15 min at room temperature. The sections were in a moist chamber and incubated overnight at 4uC with primary antibodies for fascin-1 (mMAb IM20, 1:200; Vector Laboratories), ezrin (mMAb 3C12, 1:100) and Paxillin (mMAb 5H11, 1:100; both Thermo Fisher Scientific) and washed 3 times with PBST (containing PBS and 1% Tween20) for 5 min each. Sections were incubated with the secondary antibody from the Max VisionTM HRP-Polymer anti-Mouse IHC kit (MaxinBio) for 15 min at room temperature, then washed 3 times for 5 min with PBST. The DAB substrate detection system was used (Vector Laboratories, Inc.). Counterstaining involved hematoxylin, then sections were dehydrated and mounted with coverslips. Evaluation and Scoring of Immunohistochemistry Each immunohistochemically stained sample was independently evaluated by 2 pathologists with agreements,but adjuster when argued (Pathology Department, No. 1 Hospital Affiliated with Shanxi Medical University). The slides were coded and evaluators were not aware of the code, to avoid the observer bias. All images were captured by use of the MShot Digital CCD Transducer Imaging System (Microshot Technology Co.). To determine the expression pattern, at least 10 high-power fields (40610) and at least 100 cancer cells per field were selected randomly for microscopy.
Squamous lung carcinoma was selected as an appropriate positive control for fascin-1, ezrin and paxillin [16,31,37]. Negative controls were obtained by substituting the primary antibody with PBS.

Statistical Analysis
Statistical analysis involved use of SPSS v19.0 (SPSS Inc., Chicago, IL). For qRT-PCR, the relative mRNA level was analyzed by the 2 2DDCT method [41]. The relative mRNA or     protein fold change in level between total LSCC and total ANM samples involved the Mann-Whitney or Student t test.
The association of immunohistochemical status and clinicopathological variables was assessed by Pearson's chi-square or Fisher's exact test and that of fascin-1, ezrin and paxillin immunohistochemical staining and clinical and histopathological variables by Mann-Whitney or Kruskal-Wallis test. The correlation between the expression of the 3 proteins was assessed by Kendall's tab-b correlation analysis. Overall survival was defined as the interval from the date of surgery to death from laryngeal carcinoma or date of last contact. Survival was analyzed by the Kaplan-Meier method with the Log rank test. Multivariate analyses involved use of Cox proportional hazards models. For all tests, a two-tailed P#0.05 was considered significant. Data display involved use of GraphPad Prism v6.0 Demo (San Diego, CA).

Results
A summary of study population and clinicopathological variables is in Table 1.

Immunohistochemistry
Fascin-1, ezrin or paxillin was mainly localized in the cytoplasm of tumor cells (Figures 3, 4 and 5). Immunoreactivity was uniform in the central tumor areas but was increased in the invasion front of the tumor. However, cytoplasmic staining was usually lost in keratinized cells or keratin pearls. Staining for the 3 proteins was positive in the cytoplasm of inflammatory cells (mainly lymphocytes). Staining for fascin-1 and paxillin but not ezrin was found in vascular endothelial cells. The expression was high for fascin-1 in 66.7% cases (144 of 216), for ezrin in 59.7% cases (129 of 216), and for paxillin was high in 69.0% cases (149 of 216).

Association of Protein Expression and Clinicopathological Data
The data on the association between the expression ratio distribution of 3 cytoskeleton proteins and clinicopathological characteristics is listed in Table 2 and the correlation is shown in Table 3.
The expression of fascin-1 (P = 0.023) and ezrin (P = 0.018) was higher for smokers (132 cases) than non-smokers (84 cases). Nevertheless, only fascin-1 expression intensity was higher for smokers than non-smokers (mean rank 115.11 vs. 98.11, P = 0.019). The expression of the 3 proteins did not differ by sex or metastasis classification.

Follow-up and Survival Analysis
The last follow up was carried out in December 2011. The median age of the 216 patients at the time of surgery was 59.0 years (range 18-82 years). The median follow-up was 65 months (range 4-126 months). The data of follow-up was recorded by use of ''The otolaryngology follow-up computer program'' which was programed by our department. Patients did not undergo radiotherapy, chemotherapy or biotherapy after surgery. Sixteen patients were lost to follow-up. In total, 60 cancer-related deaths occurred, including recurrence in situ (37 patients

Discussion
Our results demonstrate significantly greater expression of fascin-1, ezrin and paxillin in LSCC than ANM tissues, which agrees with observations in ESCC [15,34], gastric cancer [42], and NSCLC [31,37]. Fascin-1, ezrin or paxillin might promote the malignant progression of LSCC; therefore, we investigated whether fascin-1, ezrin or paxillin levels could be potential molecular markers to determine LSCC prognosis. We found that high expression of fascin-1, ezrin and paxillin was associated with poor prognosis.
Fascin-1, an actin cross-linking protein, is found in the core actin bundles of cell-surface spikes and projections, which are implicated in cell motility. Goncharuk et al. [43] proposed that fascin-1 may play a role in squamous epithelium malignancy. In addition, overexpression of fascin-1 may be a useful prognostic marker in LSCC to improve the early identification of patients with a potentially highly aggressive clinical course [44]. Likewise, we observed a high protein level of fascin-1 in 66.7% of LSCC patients. We found a positive correlation between high expression of fascin-1 and advanced tumor stage (T3+T4), poor cancer differentiation, N+ cancer, and advanced clinical stage (III+IV). This observation was in line with the work of Durmaz et al. [18] and Zou et al. [19] for LSCC and other researchers for a number of other malignant neoplasms [12,14].
In the present study, the expression of fascin-1 was greater in supraglottic than glottic LSCC. Supraglottic LSCC represented 33 cases (60.0%) of cervical lymph node metastasis, which decreased to 20 cases (36.4%) with glottic LSCC. A rich lymphatic drainage is one salient feature of laryngeal anatomy in the supraglottic region, where cervical lymph node metastasis often occurs in LSCC [5]. The different expression patterns of fascin-1 in primary sites of LSCC may be due to differential incidences of cervical lymph node metastasis.
For LSCC patients in this study, fascin-1 expression status was a significant independent predictor of survival time after surgery, which was similar to reports by Zou et al. [19] showing high fascin-1 expression correlated with poor disease-free survival (DFS) and an independent predictor of DFS in LSCC. These results were not observed by Durmaz et al. [18]. In other malignant neoplasms, including gastric cancer [13], oral squamous cell carcinoma [14], ESCC [15], NSCLC [16], and breast carcinoma [17], high expression of fascin-1 was correlated with poor patient prognosis and/or decreased DFS, which supports our findings. Jawhari et al. [11] indicated a possible role for fascin-1 in the development of colonic carcinoma through augmented cell motility and enhanced metastatic potential with downregulation of hsa-mir-145/143 [45]. The in vitro interference in breast cancer cell lines demonstrated a central role for fascin-1 in regulating the cell morphologic, migration and invasion potential by modulating several metastasis-associated genes. Fascin-1 could downregulate the expression and nuclear translocation of key metastasis suppressor proteins such as breast cancer metastasis suppressor-1 and upregulate NF-kappa B activity. Importantly, fascin-1 upregulated other proteins that are critical for the execution of metastasis, such as urokinase-type plasminogen activator and matrix metalloproteases 2 and 9 [17]. These might be potential or similar mechanisms that could explain why fascin-1 overexpression in the cytoplasm of LSCC cells leads to a more clinically aggressive course.
Interestingly, the expression of fascin-1 was higher in smokers than non-smokers. Smoking is an important etiological factor of LSCC predisposition [46]. Cotinine, a main nicotine metabolite, causes cytoskeletal alterations, which influences cell size and shape and induces a mitotic catastrophe phenotype in NSCLC cells [47]. Fascin-1 and smoking may be cofactors in the development of LSCC, which warrants further research.
Ezrin, a member of the ERM family of proteins, is a key signaling molecule that regulates cell survival, adhesion, migration and invasion. It also plays a role in binding plasma membrane proteins with the actin cytoskeleton, thereby providing an intracellular scaffold for the formation of specialized membrane domains that facilitate signal transduction through a number of growth factor receptors and adhesion molecules [48]. Ezrin is associated with malignant progression and metastasis in a variety of human neoplasias [25,26,28,30,23]. We found 59.7% of LSCC sections with high ezrin expression, which was similar to findings by Xie et al. [28] for ESCC, Wang et al. [49] for salivary gland adenoid cystic carcinoma, and reports of HNSCC [23]. Zhang et al. [31] found positive expression of ezrin common in stage III NSCLC with lymph node metastasis. These data corroborated the present results, which confirm that high expression of ezrin is related to advanced clinical stage (III+IV) and cervical lymph node metastasis. Meanwhile, Wang et al. indicated that increased ezrin expression was significantly correlated with high lymph nodal metastatic rate in nasopharyngeal cancer [50]. An analysis of hepatitis B-related hepatocellular carcinoma showed that ezrin overexpression was independently associated with poor differentiation and invasion [51], which is consistent with our results. Akisawa et al. [24] investigated the expression of ezrin in 16 pancreatic adenocarcinoma cell lines with different metastatic potential. Two lines, S2-CP9 and S2-VP10, with high metastatic potential, showed significantly higher mRNA and protein levels of ezrin than did other cell lines. Ezrin upregulation was confirmed and considered an important regulator in lung cancer bone metastasis, which was induced by transforming growth factor b [30]. These results suggest a strong correlation between high ezrin expression in LSCC and metastatic potential. Furthermore, we found a high expression of ezrin associated with advanced T stage (T3+T4) and age $59 years. These data suggest a key role of ezrin in late LSCC progression and metastasis, which needs further investigation. A high expression of ezrin was related to poor outcome, which supported findings in HNSCC [23], ESCC [28], NSCLC [31], salivary gland adenoid cystic carcinoma [49], nasopharyngeal cancer [50], and endometrioid carcinoma [52]. We found that ezrin expression was an independent prognostic factor of LSCC. Although this was not observed in HNSCC [23], it remained a predictor in other multivariate analyses [28,31,49,50,52].
Paxillin is an adapter for scaffold protein regulation, signaling and focal adhesion assembly, with the primary function being providing multiple docking sites on the plasma membrane for an array of signaling and structural proteins [53], and controlling the dynamic changes during cell adhesion and cytoskeletal reorganization [54]. It has been linked to many malignancies [33,34,35,36,37,38,39] except LSCC. Paxillin was highly expressed in 69.0% of our cases, which was similar to the proportion found for HCC [36] and urothelial bladder tumor [55]. High expression of paxillin was positively correlated with poor tumor differentiation, cervical lymph node metastasis and advanced clinical stage (III+IV). In HCC, positive paxillin expression was associated with low differentiation and extra-hepatic metastasis [36]. Upregulated paxillin mRNA and protein expression was prevalent in stage III NSCLC [37]. These findings agree with ours for LSCC. However, Li et al. [35] suggested that positive paxillin expression in ESCC, although significantly higher than normal squamous tissue, had no significant relation to clinicopathological variables and in predicting ESCC prognosis. Overall, we observed  a correlation between paxillin overexpression and LSCC aggressiveness. High paxillin expression was a predictor of poor survival and relapse-free survival but also an independent predictor of NSCLC [56]. We found that high expression of paxillin indicated poor postoperative LSCC survival. However, paxillin was not an independent predictor of LSCC, which agrees with observations in HCC [36]. Positive ezrin and paxillin protein expression was seen in 99% and 93.7%, respectively, of urothelial tumors; downregulation of ezrin and paxillin in urothelial bladder tumors was associated with aggressive tumor features and invasiveness [55]. Different tumor origin, extracellular environments and engagement of different signaling partners may account for these discrepant results. Nonetheless, the mechanism of paxillin in LSCC is still not well understood. Two studies of HNSCC cell lines (UMSCC-1 and SCC 25) indicated that paxillin overexpression contributed to increasing adhesion ability [57] and that paxillin downregulation played a role in suppressing tumor cell proliferation and angiogenesis [39]. Enhanced Ki-67 protein, a cellular marker for proliferation, increased the growth and proliferation of lung cancer cell H522 in vivo with overexpression of paxillin [58]. Increased mutant A127T paxillin expression conferred cell growth, oncogenic transformation, and tumor growth and invasion in a nude mouse model. Moreover, a recent report indicated that paxillin upregulation in response to miR-218 reduction enhanced tumor growth and metastasis in lung cancer [37]. Therefore, paxillin may play an essential role in LSCC, acting as an oncogene and driving the LSCC malignant progression. Cytoskeleton-associated proteins regulate polarity, differentiation, proliferation, migration and invasion of neoplastic cells by their intimate association with the actin cytoskeletal network, a complex mechanism [9]. Their interactions could contribute to the signal transduction, which is involved in the transforming growth factor b, b-catenin-TCF, MARK, and Twist pathways [9,12,28,33,30], between cellular surface projections (microvilli and membrane ruffle) and cytoplasmic microfilaments, which might be the structural elements responsible for the migratory invasive properties of LSCC [59]. Fascin-1, ezrin and paxillin are regulated by microRNAs and take part in the modulation of malignant progression in carcinoma [29,37,45]. Our study demonstrated a positive correlation among the protein levels of fascin-1, ezrin and paxillin, so they may have a synergistic effect during LSCC invasion. Further studies are needed to understand the mechanism of regulation among these proteins, as well as the involvement of growth factors, transcription factors and micro-RNAs in LSCC.
In summary, we evaluated the expression patterns of the cytoskeleton-associated proteins fascin-1, ezrin and paxillin in tissue and paraffin sections of LSCC. The 3 proteins may be biomarkers for the LSCC malignant process and effective prognostic markers for poor outcome of patients after surgery. Notably, fascin-1 and ezrin may be independent prognostic factors for LSCC after surgery. This study may provide promising new molecular therapeutic targets, a novel strategy of individual therapeutic or useful biomarkers for prognosis of LSCC. However further studies, such as use of short hairpin RNA to downregulate the expression of the factors, either individually or in combination, in human LSCC cell lines are needed. As well, study is needed to determine any changes in cell behavior (proliferation, cell cycle, migration, invasion, metastasis) in vitro and in vivo, to confirm the findings.