Roles of ezrin in the growth and invasiveness of esophageal squamous carcinoma cells

Authors

  • Jian-Jun Xie,

    1. Department of Biochemistry and Molecular Biology, Medical College of Shantou University, Shantou, People's Republic of China
    Search for more papers by this author
  • Li-Yan Xu,

    Corresponding author
    1. Department of Pathology, The Key Immunopathology Laboratory of Guangdong Province, Medical College of Shantou University, Shantou, People's Republic of China
    • Institute of Oncologic Pathology, The Key Immunopathology Laboratory of Guangdong Province, Medical College of Shantou University, Shantou 515041, People's Republic of China
    Search for more papers by this author
    • Fax: +86-754-88900847

  • Yang-Min Xie,

    1. Department of Experimental Animal Center, Medical College of Shantou University, Shantou, People's Republic of China
    Search for more papers by this author
  • Hai-Hua Zhang,

    1. Department of Biochemistry and Molecular Biology, Medical College of Shantou University, Shantou, People's Republic of China
    Search for more papers by this author
  • Wei-Jia Cai,

    1. Department of Pathology, The Key Immunopathology Laboratory of Guangdong Province, Medical College of Shantou University, Shantou, People's Republic of China
    Search for more papers by this author
  • Fei Zhou,

    1. Department of Biochemistry and Molecular Biology, Medical College of Shantou University, Shantou, People's Republic of China
    Search for more papers by this author
  • Zhong-Ying Shen,

    1. Department of Pathology, The Key Immunopathology Laboratory of Guangdong Province, Medical College of Shantou University, Shantou, People's Republic of China
    Search for more papers by this author
  • En-Min Li

    Corresponding author
    1. Department of Biochemistry and Molecular Biology, Medical College of Shantou University, Shantou, People's Republic of China
    • Department of Biochemistry and Molecular Biology, Medical College of Shantou University, Shantou 515041, People's Republic of China
    Search for more papers by this author
    • Fax: +86-754-88900847


Abstract

Ezrin, which crosslinks the cytoskeleton and plasma membrane, is involved in the growth and metastatic potential of cancer cells. Ezrin expression in esophageal squamous cell carcinoma (ESCC) was described recently, but its roles and the underlying mechanism(s) remain unclear. In our study, we first showed that ezrin in ESCC cell is expressed in the nucleus as well as in the cytoplasm and plasma membrane. Then, by using RNAi, we revealed that interference of ezrin expression suppressed the growth, adhesion and invasiveness of ESCC cells. Tumorigenesis experiments revealed that ezrin may directly regulate tumor formation in vivo. To explore the molecular mechanisms through which ezrin contributes to the proliferation and invasiveness of ESCC cells, we used cDNA microarrays to analyze ezrin knockdown cells and the control cells; of 39,000 genes examined, 297 were differentially expressed upon ezrin knockdown, including some proliferation- and invasiveness-related genes such as ATF3, CTGF and CYR61. Furthermore, pathway analysis showed that ezrin knockdown led to decreased activation of the TGF-β and MAPK pathways, and ezrin-mediated cell invasiveness alteration was dependent on the activation of these pathways. Finally, immunohistochemical staining on 80 ESCC specimens and 50 normal esophageal mucosae revealed that the expression levels of 3 altered genes involved in the regulation of cell proliferation and tumor metastasis, including CTGF, CYR61 and ATF3, were altered in ESCCs, and their expression pattern correlated with ezrin expression. Taken together, we propose that ezrin might function in the growth and invasiveness of ESCC cells through the MAPK and TGF-β pathways. © 2008 Wiley-Liss, Inc.

Esophageal squamous cell carcinoma (ESCC) is one of the most fatal malignancies worldwide. One of the reasons for its poor prognosis is that ESCC exhibits extensive local invasion or regional lymph node metastasis even at initial diagnosis.1, 2 Metastatic tumors are often refractory or only partially sensitive to current therapeutic strategies and are the primary cause of cancer-related mortality.3 Therefore, a better understanding of tumor dissemination and growth is paramount, and identification of the genes that are crucial for metastatic dissemination is of great interest not only for a basic understanding of the molecular and cellular processes involved but also to provide potential new therapeutic targets.

Certain genes, MTA1 and Fascin,4–6 have been proven to be involved in ESCC invasion. Our study focuses on ezrin, which was initially isolated as a cytoskeletal component of intestinal microvilli and a substrate for tyrosine kinase.7 As a member of the ezrin–radixin–moesin (ERM) protein family, ezrin acts both as a linker between the actin cytoskeleton and plasma membrane proteins and as a signal transducer in responses involving cytoskeletal remodeling.7 Furthermore, ezrin is present in the nucleus of several cell types and is involved in transcriptional regulation.8, 9

Ezrin is expressed in several human epithelial tumors,10–15 and its overexpression in some tumor tissues or cell lines is required for metastasis,11 although little direct evidence has been collected in this regard. Moreover, little is known about ezrin function in tumor formation. Some work has been done on the roles of ezrin in human ESCC. We found that upregulated ezrin expression is an important factor possibly associated with the invasive phenotype of malignantly transformed esophageal epithelial cells.16 Our more recent study on esophageal epithelial tissues showed that ezrin has a tendency to translocate from the plasma membrane to the cytoplasm in the progression from normal epithelium to invasive carcinoma of the esophagus.17 Nonetheless, the precise functions of ezrin in ESCC and the underlying mechanism(s) are not clearly understood.

In our study, we show that ezrin is not only constitutively found in membranes and the cytoplasm but also in the nucleus of ESCC cells. By using RNA interference (RNAi), we also stably suppressed the expression of ezrin in EC109 cells, an ESCC line, and demonstrated that ezrin knockdown suppresses the growth, adhesion and invasiveness of cells in vitro. Tumorigenesis experiments in nude mice also revealed that ezrin might regulate tumor formation in vivo. Furthermore, cDNA microarray analyses indicated that certain proliferation- and invasiveness-related genes are involved in ezrin function. Finally, we report that ezrin likely influences the growth and invasiveness of ESCC cells through MAPK and TGF-β pathways, or by affecting expressions of certain cell proliferation- or invasiveness-related genes.

Abbreviations:

ERM, ezrin-radixin-moesin; ESCC, esophageal squamous cell carcinoma; FCS, fetal calf serum; MMP-2, metalloprotease 2; MMP-9, metalloprotease 9; RNAi, RNA interference; siRNA, small interfering RNA.

Material and methods

Cell lines and RNAi

Human ESCC cell lines (EC109, EC8712, EC171 and SHEEC), HeLa cells and SHEE cells (a kind of immortal embryonic esophageal epithelium established by our laboratory18) were maintained in 199 medium (Invitrogen, Carlsbad, CA) containing 10% fetal calf serum (FCS) in an atmosphere of 5% CO2 at 37°C. The mammalian expression vector, pSUPER.neo.circular (OligoEngine, Seattle, WA), was used for small interfering RNA (siRNA) expression in EC109 cells. Briefly, 2 siRNA pairs were synthesized; one pair encoded ezrin nucleotides 274–292 (PSE1, TCCACTATGTGGATAATAA) and the second pair encoded nucleotides 265–283 (PSE2, ACTTTGGCCTCCACTATGT). The pSUPER.neo vector of nonspecific siRNA was used as a negative control (PSC). The siRNA expression plasmids were transfected into EC109 cells using FuGENE 6 transfection reagent (Roche, Indianapolis, IN) according to the manufacturer's instructions. G418 (400 μg/ml, Calbiochem, Darmstadt, Germany) was added to the culture medium after 24 hr. Stable G418-resistant clones were obtained in 7–9 days. The expanded cells were then used for subsequent studies. Cells transfected with PSE1, PSE2 or PSC were designated PSE1 cells, PSE2 cells or PSC cells, respectively.

Western blotting

Total cell lysates were prepared in RIPA buffer [50 mM Tris.HCl, pH 8.0/150 mM NaCl/1% (vol/vol) Nonidet P-40/0.5% (wt/vol) sodium desoxycholate/0.1% (wt/vol) SDS] containing protease inhibitors, and cellular lysates were cleared by centrifugation at 14,000 g for 10 min at 4°C. Nuclear proteins and cytoplasmic proteins were extracted from cells according to the instructions provided with the nuclear extraction kit (Active Motif, Tokyo, Japan). Equal amounts of protein were separated by SDS-PAGE and transferred to a PVDF membrane (Millipore, Billerica, MA). The membrane was incubated in blocking buffer (PBS containing 0.1% Tween 20 and 5% nonfat milk) for 1 hr at room temperature followed by the addition of the primary antibody for 2 hr at room temperature. Then, the membrane was rinse 3 times with PBS/Tween 20 and incubated with anti-mouse or anti-rabbit antibody for 2 hr at room temperature. Immunoreactive bands were revealed by Western blotting luminol reagent (Santa Cruz Biotechnology, Santa Cruz, CA).

Fluorescence staining

The staining procedure was carried out as described.6 After being fixed with 100% methanol at −10°C for 15 min, cells were incubated with blocking buffer (goat serum) for 20 min and incubated with primary antibody overnight at 4°C followed by incubation with FITC-labeled secondary antibody for 30 min at 37°C. Finally, the nuclei were counterstained using 4′,6-diamidino-2-phenylindole (DAPI, Sigma, St. Louis, MO) and view with a fluorescence microscope (Olympus, CKX41).

Antibodies

Two ezrin antibodies were used in our study: mouse antiezrin (MS-661, Lab Vision, Fremont, CA) and mouse anti-ezrin (sc-58758, Santa Cruz, CA). Antibodies of mouse anti-β-tubulin, mouse antinucleoporin p62, goat anti-Smad2/3, rabbit anti-p-Smad2/3, rabbit anti-ERK1/2 and mouse anti-p-ERK1/2 were purchased from Santa Cruz Biotechnology (Santa Cruz). Antibodies of mouse anti-CTGF and mouse anti-Syndecan-2 were from R&D (Minneapolis, MN). Other antibodies including rabbit anti-ATF3 (Rockland, Gilbertsville, PA), mouse anti-β-actin (Sigma, St. Louis, MO) and rabbit anti-CYR61 (Novus, Littleton, CO) were also used. Goat anti-mouse, goat anti-rabbit and donkey anti-goat peroxidase-conjugated secondary antibodies were bought from Santa Cruz Biotechnology. For fluorescence staining, goat anti-mouse FITC-labeled secondary antibody (KPL, Gaithersburg, MD) was employed. Information of these antibodies was listed in Supporting Information Table I.

Table I. Representative Cancer-Related Genes with Altered Relative Expression Upon Ezrin Knockdown
Probe setGene titleFold change (+,Up; −,Down)1UniGene ID
  • 1

    All changes, p < 0.05.

Cell apoptosis and proliferation-related genes
201631_s_atImmediate early response 3(IER3)+2.43Hs.76095
202644_s_atTumor necrosis factor, alpha-induced protein 3 (TNFAIP3)+2.00Hs.211600
212143_s_atInsulin-like growth factor binding protein 3(IGFBP3)+2.93Hs.450230
210775_x_atCaspase 9, apoptosis-related cysteine peptidase (CASP9)+2.22Hs.329502
218031_s_atCheckpoint suppressor 1(CHES1)+2.21Hs.434286
205992_s_atInterleukin 15(IL15)+3.12Hs.311958
230250_atProtein tyrosine phosphatase, receptor type, B(PTPRB)+3.21Hs.434375
209937_atTransmembrane 4 L six family member 4(TM4SF4)+8.15Hs.133527
226654_atMucin 12(MUC12)−4.12Hs.489355
211124_s_atKIT ligand(KITLG)−2.08Hs.1048
213240_s_atKeratin 4(KRT4)−6.7His.371139
Cell motility and invasiveness-related genes
201389_atIntegrin, alpha 5 (ITGA5)+2.00Hs.505654
200632_s_atN-myc downstream regulated gene 1(NDRG1)+2.74Hs.372914
205206_atKallmann syndrome 1 sequence(KAL1)+2.93Hs.521869
201042_atTransglutaminase 2(TGM2)+2.14Hs.517033
241769_atIntegrin, alpha V(ITGAV)+2.52Hs.436873
202997_s_atLysyl oxidase-like 2(LOXL2)+4.88Hs.116479
1554980_a_atActivating transcription factor 3(ATF3)+2.17Hs.460
201289_atCysteine-rich, angiogenic inducer 61(CYR61)−2.00Hs.8867
212154_atSyndecan 2(SDC2)−2.21Hs.1501
209101_atConnective tissue growth factor(CTGF)−4.32Hs.410037
212070_atG protein-coupled receptor 56(GPR56)−2.00Hs.513633
225275_atEGF-like repeats and discoidin I-like domains 3(EDIL3)−3.05Hs.482730
Transcription-related genes
203543_s_atKruppel-like factor 9 (KLF9)+2.22Hs.150557
221086_s_atzinc finger protein 312 (ZNF312)+21.00Hs.241523
209189_atv-fos FBJ murine osteosarcoma viral oncogene homolog (FOS)+3.30Hs.25647
228964_atPR domain containing 1 (PRDM1)+2.15Hs.436023
236659_x_atZinc finger protein 277 (ZNF277)−10.1Hs.489722
Signal transduction-related genes
202284_s_atCyclin-dependent kinase inhibitor 1A (CDKN1A)+2.05Hs.370771
219764_atFrizzled homolog 10 (FZD10)+2.30Hs.31664
1569263_atCasein kinase 1, delta (CSNK1D)+3.45Hs.477070
212070_atG protein-coupled receptor 56(GPR56)−2.00Hs.513633
1560552_a_atProtein phosphatase 3, catalytic subunit, alpha isoform (PPP3CA)−11.00Hs.435512

Cell growth study

The MTT assay and colony formation assay were used to evaluate cell growth. The MTT test was performed as described by Mosmann19 with some modifications. In brief, cells were seeded in 96-well plates (5 × 103 cells/well), and after incubation for 24, 48, 72 or 96 hr, MTT solution (5 mg/ml) was added to the medium. The formazan crystals that formed were dissolved, and absorption was measured at 490 nm with an automatic ELISA reader. In the colony formation assay, cells were plated at a density of 50 cells/well in 6-well plates. They were then moved to a cell incubator. After 30 days, the number of colony-forming cells (>50 cells) was calculated under a microscope. The data were expressed as mean ± SD.

Tumorigenesis experiments in nude mice

To detect the impact of ezrin on tumor cell proliferation in vivo, 1 × 106 siRNA-treated EC109 cells were injected subcutaneously into nude mice, with PSC-transfected cells and nontransfected EC109 cells serving as controls. These mice were bred as described.5 All animal studies complied with the Guide for the Care and Use of Laboratory Animals (NIH publication no. 86-23, revised 1985) and the current Chinese law on the protection of animals. Prior to injection, 24 nude mice (4–6 weeks old) were assigned at random to 4 groups. Time of tumorigenesis was recorded, and growth curves were generated. After 30 days of observation, the mice were sacrificed and the tumors were removed and weighed.

Cell invasiveness assay

The invasiveness was determined by an invasiveness chamber assay. Cells (1 × 105) were seeded onto the top chamber of a 24-well matrigel (BD Sciences, Franklin Lakes, NJ)-coated micropore membrane filter with 8-μm pores (Millipore, Billerica, MA), and the bottom chamber was filled with 0.6 ml of 199 medium with 10% FCS as a chemoattractant. The membranes were fixed and stained by Giemsa reagent, and the cells on the upper surface were carefully removed with a cotton swab after 24 hr. Invasiveness was quantified by counting 10 random fields under a light microscope (400×). Data obtained from 3 separate chambers were shown as mean values. For the pathway analysis, cells were pretreated with 10 ng/ml TGF-β1 (Invitrogen) or 50 ng/ml EGF (Invitrogen) for 24 hr and then used to invasiveness assay.

Chamber migration assay

Migration was evaluated using a modified Boyden chamber assay. Cell culture inserts containing polyethylene tetrephthalate were placed within a 24-well chamber containing 0.6 ml of 199 medium with 10% FCS. Cells (1 × 105) were seeded onto the inserts suspended in 0.2 ml of serum-free 199 medium. Nonmigratory cells were removed from the upper surface of the filter after incubation for 24 hr. Migrated cells were fixed and stained with Giemsa reagent. Migrating cells were quantified based on the procedure as described earlier.

Gelatin zymography

Cells were washed and cultured in serum-free 199 medium. After 24 hr, the conditioned medium from 107 cells was collected, concentrated 50-fold using a Nanosep 10K centrifugal device (Pall Corporation, East Hills, NY) and subjected to SDS-PAGE through 10% polyacrylamide gels copolymerized with 1 mg/ml gelatin (Sigma). Gels were incubated overnight at 37°C and then stained with 0.1% Coomassie blue R250.5 After destaining, gelatinolytic signals were photographed (FluorChemTMIS-8900, Alpha Innotech). The β-actin immunoblot of the cells and the coomassie staining of total protein in conditioned media were used to demonstrate that equal numbers of cells were present during the conditioning of the media.20, 21

Cell adhesion assay

The cell adhesion assay was performed as described.22 Wells of a 96-well plate were coated at room temperature overnight with 5 μg matrigel in a final volume of 50 μl. Additional uncoated wells were incubated to serve as a negative control. Cells were trypsinized from the dish, resuspended in serum-free 199 medium and 5 × 103 cells/100 μl were added to each well. The plates were incubated for 1 hr at 37°C. For quantification, the attached cells were treated with MTT solution and absorption was measured at 490/630 nm with an automatic ELISA reader.

cDNA microarray analysis

To compare the gene expression patterns between PSE1-transfected cells and PSC-transfected cells, the Affymetrix GeneChip Human genome U133 plus 2.0 arrays (Affymetrix, Santa Clara, CA) containing 39,000 known genes and expressed sequence tags were used. Total RNA of PSE1 and PSC cells was isolated, and the RNA was quantified spectrophotometrically. Double-stranded cDNA preparation, synthesis of biotin-labeled cRNA target, hybridization, washing and staining, subsequent scanning of the hybridized array and data processing were performed as specified in the Affymetrix Gene Chip Expression Analysis Technical Manual. The ratio of the geometric means of the expression intensities for each gene fragment was computed and reported in terms of the fold change (up or down). The data were deposited in the Gene Expression Omnibus (accession number: GSE6233).

Reexpression of ezrin in the ezrin-depleted cells

To generate the ezrin expression vector, the open reading frame of human ezrin cDNA was cloned into the eukaryotic expression vector pcDNA4 (Invitrogen). The pcDNA4-ezrin plasmid and pcDNA6/TR (Invitrogen) were cotransfected into PSE1 cells using SuperFect reagent (QIAGEN, Germantown, USA). Ezrin expression was induced with 1 μg/ml tetracycline (Invitrogen) and cells were harvested 24 hr later.

Treatment with TGF-β1 and EGF

EC109 and PSE1 cells were seeded into 6-well plates at 2 × 105 cells/well and cultured for 12 hr. The cells were then serum-starved overnight followed by incubation with 50 ng/ml EGF or 10 ng/ml TGF-β1, respectively. Cells were harvested after different times of treatment.

RT-PCR

Total RNA was extracted by the Trizol method (Invitrogen) and subjected to reverse transcription PCR (RT-PCR) in a two-step protocol using AMV Reverse Transcriptase and PCR Master Mix (Promega, Madison, WI). The number of cycles and annealing temperature was adjusted depending on the genes amplified. Primer sequences are available in Supporting Information Table II.

Tissue specimens and immunohistochemical staining

A total of 130 paraffin sections, including 80 ESCCs and 50 normal esophageal mucosae, were obtained for immunohistochemical staining from the Department of Pathology of Shantou University from 1998 to 2007. The sections of normal mucosa were taken from the matched distal resected margin of ESCC samples. The SuperPicTure Polymer Detection kit and Liquid Substrate kit (Invitrogen) were used to conduct immunohistochemistry according to the manufacturer's instructions. Staining was scored on the following scale: 0, no staining; 1+, minimal staining; 2+, moderate to strong staining in at least 20% of cells; 3+, strong staining in at least 50% of cells. Cases with 0 or 1+ staining were classified as negative, and cases with 2+ or 3+ staining were classified as positive.

Statistical analysis

All data are expressed as mean ± SD and were analyzed with SPSS statistic software (SPSS 12.0 by SPSS, Chicago). Comparisons between data sets were performed using the χ2 test and t-test when appropriate. p < 0.05 was considered statistically significant.

Results

Expression of ezrin in ESCC cell lines and siRNA-mediated silencing

First, we examined ezrin expression in several ESCC cell lines and HeLa cells, a tumor cell line that reportedly expresses high levels of ezrin.23 Ezrin was detected in all cell lines evaluated, and with EC109 cells expressing the highest level (Fig. 1a). Hence, EC109 cells were selected as the model for the subsequent function studies.

Figure 1.

Nuclear localization of ezrin in ESCC cells and knockdown of the ezrin gene by siRNA. Ezrin levels in whole-cell extracts (a), cytoplasmic extracts and nuclear extracts (b) were determined in various ESCC cell lines by Western blotting analysis. β-actin, β tubulin and nucleoporin p62 served as loading controls in the Western blotting. (c) Immunofluorescence analysis of ezrin expression in EC109 cells and SHEE cells (mouse antiezrin antibody, Lab Vision). Representative images showing localization of ezrin (green stain) and staining of DAPI (blue stain). Immunofluorescence with the secondary antibody alone was using as blank control. Bar, 20μm. (d) Immunofluorescence analysis of ezrin silencing by siRNA. PSE1 and PSE2 cells were ezrin-siRNA-treated EC109 cells, and PSC were EC109 cells with the nonspecific siRNA vector. PSC and untreated EC109 cells served as controls. Bar, 20 μm. (e) Ezrin gene silencing in EC109 cells was evaluated by Western blotting analysis in whole-cell extracts, cytoplasmic extracts and nuclear extracts. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Ezrin reportedly localizes in the nucleus of several kinds of cells.8, 9 To address whether this was the case in ESCC cells, we studied ezrin distribution by Western blotting and immunofluorescence microscopy. Western blotting identified ezrin in several ESCC cell lines in both the cytoplasm and nucleus (Fig. 1b). Immunofluorescence staining localized ezrin to membranes, cytoplasm and nucleus (green fluorescence) of EC109 cells, whereas only moderate ezrin staining was observed in the membrane and cytoplasm of SHEE cells, a normal esophageal epithelial cell line (Fig. 1c).

To study the function of the ezrin gene in EC109 cells, the pSUPER system was used to stably suppress ezrin expression. G418-screened EC109 cells were used to analyze the silencing effect. Immunofluorescence staining and Western blotting showed that ezrin expression decreased markedly in the siRNA-treated cells (PSE1 and PSE2) compared to controls (PSC and EC109) (Figs. 1d and 1e). Compared to Western blotting controls, ezrin expression was efficiently reduced by about 87% in PSE1 cells and by about 75% in PSE2 cells. Furthermore, ezrin expression in both the cytoplasm and nucleus was decreased in the PSE1 and PSE2 cells.

Altered growth and tumor formation of EC109 cells

The colony formation assay was used to evaluate the growth of the cells in which ezrin was silenced. After culturing in G418-containing media for 30 days, much fewer colonies were formed in the PSE1 or PSE2 cells, whereas colony formation was still obvious in control groups (Fig. 2a). To further test the negative effect of ezrin knockdown on ESCC cell growth, an MTT assay was performed and growth curves were generated (Fig. 2b). As shown by the curves, both PSE1 and PSE2 cells proliferated slower than PSC cells and the untreated EC109 cells during the first 96 hr after the cells were plated. The dramatic reduction of colony formation and growth of ezrin-silenced cells suggested that ezrin suppression may negatively regulate ESCC cell growth.

Figure 2.

Effect of ezrin gene knockdown on cell growth and tumor formation. Colony formation assay (a) and MTT assay (b) were used to evaluate the proliferation of ezrin knockdown cells. (c) Tumorigenesis in nude mice injected with PSE1, PSE2, PSC or untreated EC109 cells. The percentage of mice with tumors at each time point was recorded, and growth curves were generated. Tumors first appeared after 13 days. (d) Tumor formation in nude mice after 30 days. Left, subcutaneous tumors removed from the mice. Right, average weight of the tumors. Statistical significance was determined with a Student's t-test; *p < 0.05; **p < 0.001. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Experiments were performed on nude mice to investigate the potential correlation between ezrin expression and tumor formation in vivo. Six nude mice in each group were injected with PSE1, PSE2, PSC or untreated EC109 cells, and the tumor growth was quantitated. Tumor formation was evaluated over 28 days; tumorigenesis of the siRNA-treated cells decreased significantly when compared to the controls (Fig. 2c). Furthermore, the mean weights of the tumors from the PSE1 and PSE2 groups were less than those of the 2 control groups (Fig. 2d).

Altered migration, invasiveness and adhesion of EC109 cells

The motility of PSE1 and PSE2 cells was examined by determining their migration through a polyethylene filter in the absence of matrigel. Migration rates of the PSE1 and PSE2 cells were greatly decreased when compared to the control (Fig. 3a). Ezrin knockdown also markedly reduced cell invasive properties when compared to parental cells (Fig. 3b). Cell migration rates of PSE1 and PSE2 cells were decreased by 60 and 55%, and their cell invasiveness decreased by 75 and 60%, respectively. Zymography analysis of the conditioned media obtained from cells exhibited 2 major bands at 92 and 72 kDa that were consistent with the zymographic pattern of MMP-9 (metalloprotease 9) and MMP-2 (metalloprotease 2) proenzymes, respectively. The activities of both MMP-2 and MMP-9 of siRNA-treated cells were significantly lower than those of the controls (Fig. 3c).

Figure 3.

Role of ezrin in cellular migration, invasiveness and cell-matrix adhesion. (a) Cell migration was assessed with modified Boyden chamber inserts. (b) Cell invasiveness was examined with matrigel-coated chambers. Migrated and invasive cells were fixed and stained, and representative fields were photographed. For quantification, the cells were counted in 10 random fields under a light microscope (×400). Arrows demonstrate cells that have migrated or invaded through the membrane (c) Gelatin zymography of the conditioned medium. The β-actin immunoblot of the cells and the coomassie staining of total protein in conditioned media were used to demonstrate that equal numbers of cells were present during the conditioning of the media. (d) Cell adhesion was determined by cell adhesion assay. Adhesive cells of representative fields were photographed before treatment with MTT (×400). Statistical significance was determined with a Student's t-test; *p < 0.05; **p < 0.001.

A cell adhesion assay was conducted to assess the effect of ezrin suppression on cell adhesion. The adhesion ability of the EC109 cells decreased with ezrin knockdown when compared to the control; PSE1 cells decreased by 65% and PSE2 cells decreased by 45% (Fig. 3d).

Genes differentially expressed upon ezrin knockdown

We used cDNA microarrays to identify genes that were differentially expressed upon ezrin knockdown. Of the 39,000 examined, 297 were differentially expressed by ≥2-fold between PSE1 and PSC samples; 51 genes were downregulated and 246 genes were upregulated. The intensity of the gene expression signal from PSE1 cells relative to that from PSC cells was also determined. Forty-three genes were expressed at ≥10-fold different levels in the PSE1 set when compared to the control. An additional 31 gene fragments were expressed at 5- to 10-fold different levels, and a total of 223 gene fragments were expressed at 2- to 5-fold different levels between the 2 sample sets.

To determine whether the genes affected by ezrin knockdown were involved in specific cellular functions, those genes having altered expression in the cDNA microarrays were categorized based on their biological functions. The following cellular functions were predominantly represented in the microarray analysis of genes affected by ezrin knockdown: cell adhesion and invasiveness, cell apoptosis and proliferation, transcription and cell signaling (Table I). To confirm the microarray data, 4 of these genes were randomly selected, and their relative expression was verified by RT-PCR and Western blotting. Indeed, the expression levels of connective tissue growth factor (CTGF), cysteine-rich angiogenic inducer 61 (CYR61) and syndecan 2 (SDC2) were downregulated and activating transcription factor 3 (ATF3) was upregulated in PSE1 and PSE2 cells when compared to PSC cells and EC109 control cells, validating the microarray data (Fig. 4a). Moreover, Western blotting and fluorescence staining results were consistent with RT-PCR results in terms of differential gene expression (Figs. 4b and 4c). These data suggested that multiple genes are involved in the ezrin function in ESCC cells.

Figure 4.

Verification of the expression of certain differentially expressed genes. Differential gene expression was determined by RT-PCR (a), Western blotting (b) and fluorescence staining (c). (d) Ezrin was reexpressed in the PSE1 cells and expressions of CTGF, CYR61 and ATF3 were also determined by RT-PCR. β-actin and GAPDH served as a loading control of Western blotting and RT-PCR, respectively. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

To further prove that ezrin is the cause of the documented changes in expression, tetracycline-induced reexpression of ezrin was carried out in the PSE1 cells. RT-PCR analysis revealed that, with the recovered ezrin expression, the levels of CTGF, CYR61 and SDC2 were increased and the expression of ATF3 was decreased, which were consistent with the microarray data (Fig. 4d).

Analysis of cellular pathways affected by ezrin knockdown

To investigate how ezrin affects the expression of certain genes, we used Western blotting to assess alterations in several pathways. The MAPK pathway is required for ezrin-mediated early metastasis,8 and therefore we evaluated the activation of the MAPK pathway in ezrin knockdown cells. Ezrin knockdown caused a decrease in the phosphorylation of ERK (p-ERK) (Fig. 5a). Further results showed that treatment with EGF could induce the activation of ERK/MAPK pathway in both PSE1 cells and EC109 cells, but the p-ERK level of PSE1 cells was lower than that of EC109 cells in each corresponding time point, indicating that the activation of this pathway was altered in the ezrin-suppressed cells (Fig. 5b).

Figure 5.

Alterations in the MAPK and TGF-β pathways upon ezrin knockdown. (a) Western blotting was used to determine the effect of ezrin knockdown on MAPK pathway. (b) Activation of MAPK pathway in EC109 cells and PSE1 cells after different times of EGF treatment. (c) The effect of ezrin knockdown on TGF-β pathway was determined by Western blotting. (d) Expressions of p-Smad2/3 in EC109 cells and PSE1 cells after different times of TGF-β1 treatment. Invasiveness of EC109 and PSE1 cells were addressed under control conditions or pretreated with EGF (e) and TGF-β1 (f). In (b) and (d), signal intensities for the expression of p-ERK1/2 or p-Smad2/3 were quantified by densitometric scanning, normalized by internal control (β-actin) and relative to the expression of corresponding protein in EC109 cells (0 hr).

Gene expression analysis of PSE1 and PSC cells exhibited altered expression of certain TGF-β target genes such as CTGF and CYR61.24–26 Therefore, the phosphorylation level of Smad2/3, which is a key component of the TGF-β pathway,27 was evaluated by Western blotting. The level of p-Smad2/3 was decreased with ezrin knockdown in PSE1 and PSE2 cells (Fig. 5c). Treating EC109 cells and PSE1 cells with TGF-β1 showed that expressions of p-Smad2/3, CTGF and CYR61 were increased in both cell lines in a time-dependent manner. However, the expression levels of the 3 molecules in PSE1 cells were also lower than that in EC109 cells in each corresponding time point, suggesting that ezrin knockdown might lead to the decreased activation of TGF-β pathway (Fig. 5d).

To assess whether ezrin-mediated suppression of these pathways was responsible for the decreased cell invasiveness of PSE1 cells, we tested the effects of EGF and TGF-β1 on the invasive ability of EC109 cells and PSE1 cells. By using invasiveness assay, we found that treatment with EGF or TGF-β1 not only significantly facilitated the invasiveness ability of EC109 cells but also resulted in the recovery of the invasiveness ability of PSE1 cells. However, the invasive ability of PSE1 cells after treatment of EGF or TGF-β1 did not recover to the level of untreated EC109 cells (Figs. 5e and 5f). Thus, we reasoned that ezrin-mediated cell invasiveness alteration was partially dependent on the activation of MAPK and TGF-β pathways.

Expression of CTGF, CYR61 and ATF3 in ESCC

Genes related to cell adhesion, invasiveness, apoptosis and proliferation were possible final targets that could be involved in ezrin function. Three of these genes (CTGF, CYR61 and ATF3) were selected for more detailed study. The expression pattern for each of these gene products was investigated by immunohistochemical staining of 80 ESCC specimens and 50 normal esophageal mucosae.

CTGF and CYR61 showed cytoplasmic immunostaining in both cancerous and normal mucosae of the esophagus. In ESCC samples, most of the cancer cells exhibited intense immunostaining of the 2 proteins, whereas negligible or weak positive signals were seen in the normal esophageal epithelium (Figs. 6a and 6b). Overexpression (rating of 2+ to 3+) of CTGF and CYR61 was apparent more often in ESCC than in the normal mucosa (both p < 0.001). About 79% (63/80) of the ESCC samples but only 16% (8/50) of the normal samples were positive for CTGF. For CYR61, the positive rate was 56% (43/80) in ESCCs and 18% (9/50) in normal mucosa. ATF3 immunostaining was observed in the cytoplasm and nucleus. In normal mucosae, strong positive staining for ATF3 was apparent in all epithelial layers other than the basal layer (which had negligible or weak staining). In the carcinoma portion, ATF staining was diffuse/weak in all epithelial layers (Fig. 6c). Decreased expression (0 to 1+) of ATF3 was found more frequently in ESCC than in the normal mucosa (p < 0.001). The positive rate for ATF3 was 25% (20/80) in ESCCs and 96% (48/50) in normal mucosa.

Figure 6.

Photomicrographs of CTGF (a), CYR61 (b), ATF3 (c) and ezrin (d) expression by immunohistochemical staining in ESCC (Cancer) and normal esophageal epithelium (Normal). Scale bars, 20 μm. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Ezrin expression was also assessed in these tissue sections. In normal mucosae, moderate staining for ezrin was found in the membrane and weak staining was present in the cytoplasm. In contrast, ESCC samples showed strong staining in the cytoplasm and membrane, and weak to moderate staining in nuclear (Fig. 6d). Overexpression of ezrin was found more often in ESCC than in the normal mucosa (p < 0.001). The positive rate for ezrin in ESCC was 93.8% (75/80) and 44% (22/50) in normal mucosae.

Furthermore, we identified a positive correlation between ezrin levels and the levels of CTGF (r = 0.658; p < 0.001) and CYR61 (r = 0.432; p < 0.05). We also identified a negative correlation between ezrin levels and the level of ATF3 (r = 0.643; p < 0.001) in ESCC samples. These correlations were consistent with the microarray data.

Discussion

Ezrin has been implicated in many aspects of cancer cell biology, for example, as a conduit for signals between metastasis-associated cell surface molecules and as a linker between the cytoskeleton and membrane proteins involved in cancer metastasis.11–13 Nonetheless, the exact role of ezrin in tumor progression is unknown, and the function of ezrin in tumor cell biology of ESCC has not been thoroughly investigated. To address these issues, we evaluated ezrin expression with regard to possible direct correlations with cell growth, adhesion and invasiveness of ESCC cells. We also demonstrated that ezrin might regulate the growth and invasiveness of ESCC cells through MAPK and TGF-β pathways.

Previous studies have revealed a relationship between ezrin and cell proliferation.28, 29 However, little has been reported about the effect of ezrin expression on tumor formation in vivo. Our results indicated that ezrin suppression resulted in decreased cell growth in vitro, and further experiments using nude mice also suggested a direct positive role of ezrin in tumor formation. Changes in cell growth might be key factors in regulating neoplastic progression,30 and thus our results indicate a possible role for ezrin in ESCC development. Ezrin has also been reported to have a crucial role in the dissemination of several tumors.13, 14, 31, 32 However, the expression of ezrin in these tumors suggested that ezrin has cell-specific functions in tumor invasiveness.33 Our cell invasiveness assay and chamber migration assay revealed that ezrin knockdown led to decreased cell migration and hence decreased invasiveness. In addition, ezrin also has an important role in cell adhesion and cell–cell communication.33 Our studies confirmed that the depletion of ezrin by RNAi resulted in reduced cell adhesion.

As a linker between the actin cytoskeleton and plasma membrane proteins, ezrin alteration may lead to cytoskeletal remodeling.34 Therefore, the observed effect of ezrin silencing on cell migration and invasiveness might be partially caused by changes in the cell cytoskeleton, disruption of which may enhance cell motility or by the observed decrease in activity of extracellular matrix proteases such as MMP-2 and MMP-9, which play key roles in local invasiveness.35 These mechanisms, however, have not been thoroughly investigated, and the mechanisms of the ezrin effect on tumor growth have remained unclear. We showed by cDNA microarray that the expression of several genes related to cell invasiveness and cell proliferation was altered in response to ezrin knockdown, indicating that ezrin might contribute to tumor growth and invasiveness by affecting the expression of these genes. Moreover, expression of certain transcription factors and signaling molecules, for example, Kruppel-like factor 9 (KLF9) and G protein-coupled receptor 56 (GPR56), were changed in the ezrin knockdown cells, demonstrating that ezrin may contribute to tumor biology through diverse complex mechanisms.

Ezrin is a signal transduction component, and ezrin-mediated early metastasis is partially dependent on activation of the ERK/MAPK pathway in osteosarcoma.12 Consistent with these facts, our results suggest that cells with ezrin knockdown showed decreased phosphorylation of ERK and activation of ERK/MAPK pathway might partially recover the ezrin-mediated suppression of cell invasiveness. Another important finding of our study is showing that TGF-β pathway is involved in ezrin-mediated cell alterations. We also found that ezrin knockdown affected the expression of certain putative genes of the pathway, including CYR61, CTGF. All of these genes participate in the growth or invasiveness of cancer cells,36, 37 and the protein levels of CTGF and CYR61 were altered in ESCCs, and thus these proteins might participate in tumor progression, indicating their possible involvement in ezrin function. Furthermore, the downregulation of SDC2, which has been reported to interact with ezrin,38 was found in the ezrin-depleted cells. SDC2 is also a regulator of TGF-β pathway.39 Hence, we proposed that the effect of ezrin on TGF-β pathway might be mediated by SDC2. These findings suggest that ezrin might effect the proliferation and invasiveness of ESCC cells through the activation of MAPK and TGF-β pathways. But how ezrin effects these pathways needs further study.

Ezrin has also been suggested to be localized in the nucleus; in endothelial cells, nuclear ezrin represses transcription.8, 9 In our study, ezrin was found in the nucleus as well as the plasma membrane and cytoplasm of ESCC cell line; but seldom found in the nucleus of the normal esophageal epithelial cell line. Another intriguing finding is that nuclear expression of ezrin was found more frequent in ESCC tissue than in normal tissue (data not showed), indicating that nuclear ezrin may also be involved in the progression of ESCC. But this should be confirmed by larger scale studies of human carcinoma specimens. Moreover, we showed that nuclear expression of ezrin decreased sharply in siRNA-treated cells. We postulate that the altered expression of genes caused by ezrin knockdown may be mediated, at least partially, by the decreased levels of ezrin in the nucleus. Studies are underway to clarify the mechanisms by which ezrin affects the expression of specific genes identified here.

Acknowledgements

The authors thank Dr. Jet-Fu Chiu for stimulating discussions and Dr. Shijing Fang for kind help in siRNA design.

Ancillary