Ann Surg Oncol (2010) 17:1927–1936 DOI 10.1245/s10434-010-0922-6

ORIGINAL ARTICLE – TRANSLATIONAL RESEARCH AND BIOMARKERS

Altered Expression of E-cadherin by Hepatocyte Growth Factor and Effect on the Prognosis of Nasopharyngeal Carcinoma Li-qiong Xie, MPhil1, Li-juan Bian, MPhil1,2, Zhi Li, PhD1, Yang Li, PhD1, Zhi-xun Li, MBBS1, and Bin Li, MBBS1 1 2

Department of Pathology, The First Affiliated Hospital, Sun Yat-Sen University, Guangzhou, China; Department of Pathology, Second Affiliated Hospital, Sun Yat-Sen University, Guangzhou, China

ABSTRACT Background and Purpose. Hepatocyte growth factor (HGF)-related E-cadherin expression and prognosis of patients with nasopharyngeal carcinoma (NPC) are not fully understood. This study investigated HGF-induced altered expression of E-cadherin and the relationship between prognosis and modulation of E-cadherin by HGF in NPC. Methods. 135 cases of NPC were collected, and expression of HGF, c-Met, and E-cadherin in tissue microarray was evaluated by immunohistochemical staining. Correlation between immunostainings and clinicopathological parameters, as well as the follow-up data of patients, was analyzed statistically. The association and alteration of Ecadherin by HGF treatment in NPC cell lines were evaluated by immunocytochemical staining, Western blot, and invasion assay. Results. Both high HGF expression in tumor cells (62.9%, 85/135 cases) and nonmembranous E-cadherin expression (61.5%, 83/135 cases) were significantly associated with advanced clinical stage, lymph node metastasis, and worse prognosis of NPC patients. However, only abnormal Ecadherin expression (P = 0.001) and lymph node metastasis (P = 0.004) emerged as strong independent prognostic factors for overall survival of NPC patients. In vitro, exogenous HGF decreased and internalized E-cadherin expression from cell membrane to cytoplasm, with obvious cellular morphological change. HGF-treated NPC

L. Xie and L. Bian are joint first co-authors and made an equal contribution to this work. Ó Society of Surgical Oncology 2010 First Received: 17 November 2009; Published Online: 4 February 2010 Z. Li, PhD e-mail: [email protected]

cells exhibited significantly enhanced invasive ability in Matrigel matrix-coated Transwell chamber assay. Conclusion. HGF may contribute to cell invasion in NPC by modulating E-cadherin-mediated cell–cell adhesion through downregulation and internalization of E-cadherin. Altered expression of E-cadherin by HGF is a valuable predictor for prognostic evaluation of NPC patients.

Nasopharyngeal carcinoma (NPC) is a malignant tumor that is well known to be associated with Epstein– Barr virus (EBV) and that has a high incidence in southern China, Southeast Asia, Alaska, and North Africa.1 Unlike other head and neck cancer, NPC is characterized by aggressive invasion and metastasis, which is the major cause for poor prognosis of this tumor.2 Although the combination of radiotherapy and adjuvant chemotherapy has become standard treatment for NPC in China, the 5-year survival rate is only about 50–60% because of the frequency of distant metastasis and local recurrence, as well as long-term secondary effects of radiotherapy and chemotherapy. More effective therapies and more precise prognostic prediction for NPC are thus needed. Although various studies on NPC have been preformed recently to try to identify the mechanism of invasion and metastasis of this tumor, the precise molecular mechanism remains to be clarified. In fact, it is unknown that whether or not certain molecules get involved in invasion and metastasis of NPC, and influence prognosis of NPC patients. Hepatocyte growth factor (HGF), a mesenchymal- or stromal-derived multifunctional growth factor, can promote proliferation, motility, morphogenesis, and angiogenesis in various types of cells.3 HGF plays an important role in the development and progression of tumor. In particular, it promotes tumor metastasis by stimulating motility and invasion through mechanisms

1928

probably including decreased cell–cell adhesion, enhanced cell motility, and activated matrix metalloproteases (MMPs).4–6 The biological effects of HGF on cells are transduced by activating its transmembrane receptor tyrosine kinase c-Met. On HGF binding, c-Met undergoes dimerization and autophosphorylation on tyrosine residues, and activates intracellular signaling pathways.7 Recent studies have revealed that activation of HGF and c-Met promotes tumor progression and metastasis, consequently leading to poor prognosis in numerous human cancers, including NPC.8 Our previous study also demonstrated that expression of HGF and c-Met was higher in NPC than in normal epithelial lining of nasopharynx. Moreover, increased HGF expression in neoplastic NPC cells was closely related to lymph node metastasis and clinical stage of NPC.9 E-cadherin is a transmembrane protein with five extracellular domains and a cytoplasmic domain that interacts with catenins to connect the complex to the cytoskeleton.10 It has been revealed that E-cadherin is a principal component of adhesive junctions in polarized epithelial cells.11 Decreased or abrogated E-cadherin expression is frequently associated with metastatic cancer cells in vitro and in vivo. Cell-surface E-cadherin can be endocytosed and recycled into cell membrane, which thus plays important roles in regulating cadherin-based adhesion.12 Direct and indirect cadherin interactions with a wide range of receptor tyrosine kinases and their ligands, including the HGF and Met receptor, have been documented. Activation of Met receptor resulting in downregulation of cadherin-mediated cell–cell adhesion has been reported.13 Also, HGF expression is associated with E-cadherin expression in gastric cancer cells.14 Moreover, HGF induces ectopic E-cadherin expression from cell–cell contact site to cytoplasm and has also been considered to affect invasion and metastasis of prostate cancer cells.15 However, the effect of HGF on E-cadherin expression in NPC is still unknown. The localization of E-cadherin expression in NPC tumor cells and its relationship with invasion and metastasis, as well as prognosis of NPC, have not been examined. In the present study, we firstly studied the correlation between HGF, c-Met, and E-cadherin in NPC by tissue microarray. Our results showed strong correlation between HGF expression and abnormal E-cadherin expression in tumor cells, which also influenced the prognosis of NPC patients. Furthermore, our results showed that HGF decreased E-cadherin expression and induced cytoplasmic E-cadherin expression in NPC tumor cells in vitro, resulting in invasion into a collagen membrane. We postulate that alteration of E-cadherin expression in NPC cells by HGF may be involved in the processes of invasion and metastasis of NPC, and consequently influences the prognosis of NPC.

L. Xie et al.

MATERIALS AND METHODS NPC Specimens and Clinicopathological Findings Archival formalin-fixed, paraffin-embedded specimens from 135 primary NPC patients before treatment during 1999–2007 in the first Affiliated Hospital, and Cancer Center of Sun Yat-sen University (Guangzhou, China) were collected. The patients were 89 males and 46 females with median age of 49 years (range 21–73 years). According to World Health Organization (WHO) histological classification, 117 patients were diagnosed as having undifferentiated nonkeratinized carcinoma (UNKC) and 18 were diagnosed as having differentiated nonkeratinized carcinoma (DNKC).The patients were classified in terms of disease stage according to the criteria of the China NPC 92 staging system, with 9 in stage I, 36 in stage II, 53 in stage III, and 37 in stage IV.16 All patients were treated with standard curative radiotherapy with or without chemotherapy. Paraffin Tissue Microarray Construction In this study, the paraffin tissue microarray construction was conducted as previously described.17 Briefly, a section was cut from each paraffin block and stained with hematoxylin and eosin (H&E). Each donor block was overlaid with the corresponding H&E slide and observed by experienced pathologists. The area in the donor block for tissue microarray sampling was verified according to the H&E slide and marked. A manual tissue arrayer (Beecher Instruments Inc., USA) was used for array construction. Two representative 1.0-mm cores were removed from each donor block and transferred to a premolded recipient paraffin block with designated orientation. An additional six cores derived from tonsil, lymph node, and breast carcinoma were used as control material. Serial sections with 4lm thickness were cut from the arrayed block and mounted on aminopropyltriethoxysilane (APES)-coated glass slides and stored at 4°C for further analysis. Immunohistochemistry and Scoring The tissue array sections were subjected to immunostaining using a ChemMate Envision/HRP kit (Dako Co., Denmark). Slides were deparaffinized in xylene, rehydrated in decreasing concentrations of ethanol, and rinsed in phosphate-buffered saline. The slides were incubated with hydrogen peroxide for 10 min and then treated by microwave with 10 mM citrate buffer (pH 6.0; Sigma-Aldrich, Germany) at 5-min intervals for a total of 10 min. After blocking with normal serum for 10 min, the slides were incubated with 1:100 dilution of rabbit polyclonal HGF

Altered E-cadherin Expression by HGF in NPC

antibody (Santa Cruz Biotechnology, USA), 1:100 dilution of mouse anti-human monoclonal c-Met antibody (Santa Cruz Biotechnology, USA) or 1:100 dilution of mouse antihuman monoclonal E-cadherin antibody (Santa Cruz Biotechnology, USA) for 60 min, respectively. Slides were processed by ChemMate Envision/HRP kit for 30 min at room temperature, followed by development with diaminobenzidine (DAB) for visualization. Negative controls were included by substituting nonimmune serum for primary antibodies. Immunostaining results were evaluated and scored semiquantitatively by two pathologists without knowledge of patients’ clinical data. Evaluation of immunostaining results was calculated by double scoring system (stain intensity times stain area) as previously described. Stain intensity was scored as 0 for no staining, 1 for weak staining, 2 for moderate staining, and 3 for strong staining.18 Staining area was scored as 1 for less than 35%, 2 for 35–75%, and 3 for [75% of tumor cells. High expression of HGF or c-Met was defined when immunostaining score was C4, whereas low expression of proteins was defined for score \4. In addition, the localization of E-cadherin expression was also observed. Membranous E-cadherin expression was defined when it was solely localized in the cell membrane. When E-cadherin expression was lacking or the protein was expressed in cytoplasm, abnormal expression was defined. Cell Lines and Chemicals In this study, the human NPC cell lines CNE-1 (a welldifferentiated NPC cell line) and CNE-2 (a poorly differentiated NPC cell line) were used. Both cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM) culture medium with 10% heat-inactivated fetal calf serum (FCS) and antibiotics (50 U/ml penicillin and 100 lg/ml streptomycin, Gibco/Invitrogen) at 37°C in a humidified incubator with 5% CO2. Human recombined hepatocyte growth factor (HGF; Sigma Chemical) was dissolved in culture medium for in vitro experiments. Western Blotting Analysis and Immunocytochemical Staining In Vitro Study CNE-1 and CNE-2 cells were cultured in DMEM medium with 10% FCS until 70% cell confluency. Medium was changed to DMEM with 1% FCS (v/v), and the cells were treated with final concentration of 25 ng/ml HGF for 48 h. Cell lysates were prepared as previously described.19 Equal protein samples were subjected to 12% sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE), followed by transfer to polyvinylidene fluoride

1929

(PVDF) membrane, blocking in 5% fat-free milk, and incubation with HGF, c-Met, E-cadherin antibodies (as used in immunohistochemistry) or actin antibody (SigmaAldrich) at 4°C overnight. Detection was performed using horseradish peroxidase-conjugated secondary antibody and enhanced chemiluminescence reagents (Amersham Life Sciences, UK). Relative optical density (ROD, ratio to actin) of each blot band was quantified by using National Institutes of Health (NIH) image software (Image J 1.36b). HGF-treated and untreated CNE-1 or CNE-2 cells with serum starvation seeded on cover glasses were fixed with freshly prepared 4% paraformaldehyde in phosphate-buffered saline (pH 7.4) for 30 min at room temperature, then cells were pretreated by blocking endogenous peroxidases and were permeabilized with 0.1% Triton X-100 in 0.1% sodium citrate for 2 min. The process of primary antibody incubation and detection for HGF, c-Met, and E-cadherin was in the same way of immunohistochemical staining in the tissue array slides of NPC tissues. Cell Invasion Assay Tumor cell invasion was assayed by Transwell chamber (Becton Dickinson) with 8-lm-porosity polycarbonate filter membrane. The upper layers of membranes were precoated with 30 ll Matrigel matrix, a material that mimics the basement membrane, containing 10% DMEM, 5% 10 9 HEPES, 35% acidic collagen (rat tail), sand 50% Matrigel in each Transwell chamber, and stored at 4°C before using. Immediately before the experiment, the coated Transwell chambers were immersed into the medium (DMEM without FCS) for 2 h to rehydrate the Matrigel and to verify that all of the 8-lm pores were blocked by Matrigel matrix in case any tumor cells went through the membrane nonspecifically. CNE-1 and CNE-2 cells (500 ll, 1 9 105 cell/ml medium with 1% FCS) treated or not with 25 ng/ml HGF for 24 h were placed into the upper side of the coated and rehydrated cell chamber, respectively. This chamber was inserted into each well of a 24-well plate filled with 500 ll medium (DMEM with 5% FCS). The cells in the chamber were incubated for another 48 h at 37°C, and the cells which went through the coated membrane to the underside of the filter were counterstained by 1% crystal violet to reveal the nuclei. For each chamber, the number of invasive tumor cells was counted in five random high-power fields (HPF; 10 9 40) by reversed light microscopy. Statistical Analysis All statistical analysis was carried out by using SPSS 13.0 software for Windows. Chi-square test was used to

1930

assess HGF, c-Met, and E-cadherin expression with respect to clinicopathological characteristics. The survival curve of patients was determined by Kaplan–Meier method and Cox regression, and statistical evaluation was performed using the log-rank test. Results derived from in vitro study are presented as mean ± standard deviation (SD). Data were analyzed by one-way analysis of variance (ANOVA) with Dunnett’s post hoc test and Turkey’s post hoc test for multigroup comparisons. A P value \ 0.05 was considered statistically significant. RESULTS Correlation of HGF, c-Met, and E-cadherin Expression with Clinicopathological Parameters in NPC Positive cytoplasmic HGF staining was observed in either tumor or stromal cells, including fibroblast and lymphoid cells. High HGF expression in tumor cells was observed in 62.9% (85/135) of NPC tissues (Fig. 1a, b).

FIG. 1 Immunohistochemical staining of NPC tissue microarray. a Positive HGF staining in low-power field. b Positive cytoplasmic HGF staining in tumor cells and weak staining of stromal cells of nasopharynx. c Strong and diffuse positive c-Met staining in tumor cells. d Strong intercellular membranous E-cadherin immunoreactivity was detected in some NPC cases. However, abnormal E-cadherin expression, including intracytoplasmic expression (e) and lack of expression (f), were detected in more NPC cases. (a, immunohistochemical staining with original magnification, 9100; b–f, immunohistochemical staining with original magnification, 9400)

L. Xie et al.

High HGF expression was significantly associated with advanced clinical stage (stage III and IV) and lymph node metastasis (P \ 0.05, Table 1). Unlike HGF expression, cMet positivity was mainly localized in the cytoplasm of tumor cells of NPC. High c-Met expression was observed in 82.9% (112/135) of NPC tissues (Fig. 1c). There was a statistically significant correlation between high c-Met expression and lymph node metastasis (P \ 0.05, Table 1). High c-Met expression was also associated with high HGF expression in tumor cells. E-cadherin expression was found in 80.7% (109/135) of NPC tissues. In only 52 (38.5%) NPC tissues could Ecadherin be detected solely in the tumor cell membrane and classified as exhibiting a membranous pattern (Fig. 1d). Abnormal expression of E-cadherin, including cytoplasmic and absent expression, was found in the other 83 (61.5%) NPC tissues (Fig. 1e, f). Abnormal expression of E-cadherin was closely related to lymph node metastasis and advanced clinical stage. Furthermore, high HGF expression correlated significantly with E-cadherin expression with

Altered E-cadherin Expression by HGF in NPC

1931

TABLE 1 Correlation between expression of proteins and clinicopathological parameters of NPC patients (chi-square test) Variable

HGF expression

c-Met expression

E-cadherin expression

Low (n = 50)

High (n = 85)

Low (n = 23)

High (n = 112)

Membranous (n = 52)

Abnormal (n = 83)

\49 (n = 66)

22

44

13

53

26

40

C49 (n = 69)

28 P = 0.352

41

10 P = 0.303

59

26 P = 0.909

43

Male (n = 89)

30

59

12

77

31

58

Female (n = 46)

20

26

11

35

21

25

Age (years)

Gender

P = 0.305

P = 0.122

P = 0.222

WHO histotype DNKC (n = 18)

6

12

5

13

6

12

UNKC (n = 117)

44

73

18

99

46

71

P = 0.553

P = 0.086

P = 0.480

Tumor size T1–2 (n = 57)

19

38

8

49

19

38

T3–4 (n = 78)

31

47

15

63

33

45

P = 0.413

P = 0.384

P = 0.229

Lymph node metastasis N0 (n = 35) N1–3 (n = 100)

8 42

27 58

P = 0.018

11 12

24 88

P = 0.000

23 29

12 71

P = 0.000

Clinical stage I–II (n = 45)

10

35

9

36

25

20

III–IV (n = 90)

40

50

14

76

27

63

P = 0.007

P = 0.505

P = 0.005

c-Met expression Low (n = 23)

16

7

High (n = 112)

34

78

P = 0.003 E-cadherin expression Membranous (n = 52)

27

25

12

40

Abnormal (n = 83)

23

50

11

72

P = 0.005

abnormal pattern. However, no significant relationship between E-cadherin and c-Met expression was observed (Table 1). Association of HGF, c-Met, and E-cadherin Expression with Survival of NPC Patients In this study, 84 cases had adequate follow-up data for the final analysis, whereas 51 cases were excluded from survival analysis because the patients were lost to followup. The 84 cases were followed up from 2 to 59 months (mean 31.6 months), and 30 patients (35.7%) died of tumor during this period. Patient survival was associated with lymph node metastasis, advanced clinical stage, high HGF

P = 0.096

expression, and abnormal E-cadherin expression in tumor (Fig. 2). On multivariate analysis, only lymph node metastasis (hazard ratio, 14.246; 95% confidence interval, 2.289–88.649, P = 0.004) and abnormal E-cadherin expression of NPC tumor cells (hazard ratio, 10.231; 95% confidence interval, 2.574–40.665, P = 0.001) were independently associated with survival (Table 2). Effect of HGF on c-Met and E-cadherin Expression in NPC Cells In Vitro CNE-1 and CNE-2 NPC cells were treated with final concentration of 25 ng/ml exogenous HGF for a period of 48 h. We investigated the status of c-Met and E-cadherin in

1932

L. Xie et al.

a

b

Cumulative Survival 1.0

Cumulative Survival 1.0

0.8

0.8

0.6

0.6 Lymph node metastasis N0 N1–3

0.4 0.2

p = 0.002 0

10

20

30

40

50

0.4

Early clinical stage (stage I–II) Advanced clinical stage (stage III–IV)

0.2

p = 0.046

60

0

10

20

Months

30

40

c

d

Cumulative Survival 1.0

Cumulative Survival 1.0

0.8

0.8

0.6 0.4

0.4

0.2

p = 0.016

0.2

10

20

30

40

50

60

p = 0.001

0

Months

FIG. 2 Kaplan–Meier survival analyses of NPC patients. a Kaplan– Meier curve showing that patients with lymph node metastasis (N1–3) have lower survival rate than those without lymph node metastasis (N0). b A significant difference in survival rate was found between early clinical stage (I–II) and advanced clinical stage (III–IV). c A

TABLE 2 Cox regression model for multivariate analysis of prognostic factors in NPC

60

Expression of E-cadherin Membranous Abnormal

0.6 HGF Low expression High expression

0

50

Months

10

20

30

40

50

60

Months

significant difference in survival rate was found between NPC patients with high and low HGF expression in tumor. d Patients with membranous E-cadherin expression had significantly higher survival rate than those with abnormal E-cadherin expression in tumor

Variable

Hazard ratio 95% Confidence interval P value

Age (\49 versus C49 years)

1.575

0.734–3.381

0.243

Gender (male versus female)

1.419

0.503–3.999

0.507

WHO Histotype (DNKC versus UNKC)

0.933

0.275–3.163

0.912

Tumor size (T1–2 versus T3–4)

1.251

0.591–2.643

0.557

Lymph node metastasis (N0 versus N1–3)

14.246

2.289–88.649

0.004

Clinical stage (I–II versus III–IV)

0.599

0.218–1.645

0.321

HGF expression (low versus high)

3.127

0.415–23.557

0.268

c-Met expression (low versus high)

0.576

0.078–4.213

0.587

2.574–40.665

0.001

E-cadherin expression (membranous versus abnormal) 10.231

NPC tumor lines by immunostaining and Western blotting. Both c-Met and E-cadherin were observed in CNE-1 and CNE-2 cells. By Western blot test, expression of c-Met was significantly increased in the two NPC cell lines after 48 h of HGF treatment. The amount of E-cadherin was

decreased by HGF stimulation in both cell lines (Fig. 3). By immunostaining assay, we found the shape of some cells changed from polygonal epithelial-like to spindle mesenchymal-like, with dramatically increased expression of c-Met in the cytoplasm of HGF-treated CNE-1 cells

Altered E-cadherin Expression by HGF in NPC

1933

Effect of HGF on NPC Cell Invasion In Vitro The effect of HGF on CNE-1 and CNE-2 cell invasion was evaluated by the Matrigel matrix-coated invasion chamber method. As shown in Fig. 5a, without treatment with HGF, few CNE-1 and CNE-2 NPC cells were able to pass through the membrane to the lower chamber in 48 h; the cell counts were 35.16 ± 11.08/5HPF (CNE-1) and 29.83 ± 7.80/5HPF (CNE-2), respectively. After treatment with HGF, the number of cells that had passed through the membrane was 84.66 ± 16.99/5HPF for CNE-1 cells and 93.16 ± 28.72/5HPF for CNE-2 cells (Fig. 5b). Compared with non-HGF-treated tumor cells, a significant difference was found for both investigated NPC cell lines (Fig. 5c). FIG. 3 Effect of HGF on E-cadherin and c-Met expression in NPC cells. Exogenous HGF induced downregulation of E-cadherin and upregulation of c-Met in both CNE-1 and CNE-2 cell lines

DISCUSSION

(Fig. 4a, b). However, unlike in CNE-1 cells, in HGFtreated CNE-2 cells, the cells were obviously observed to be detached from each other and E-cadherin staining at the cell surface membrane decreased and was distributed in the cytoplasm (Fig. 4c, d).

The importance of the HGF/c-Met regulating system in cancer progression is still not fully understood. Several studies on the role of HGF in invasion and metastasis have demonstrated that HGF enhances tumor cell invasion in various human epithelial malignant tumors, including gastric cancer, lung cancer, hypopharyngeal cancer, and breast cancer.18,20–22 HGF-mediated cell migration and

FIG. 4 Immunocytochemical staining for c-Met and E-cadherin in NPC cells in vitro. a Positive cytoplasmic signal for c-Met protein was found in non-HGF-treated CNE-1 cells with polygonal shape. b After HGF treatment, the shape of some cells was found to be changed to spindle-like, and the cytoplasmic c-Met signal was significantly increased compared with non-HGF-treated CNE-1 cells.

c CNE-2 cells were cultured without exogenous HGF. E-cadherin staining was detected at the intercellular membrane, and cells were observed to adhere to each other. d After HGF treatment, cells were scattered and E-cadherin staining at cell surface membrane decreased and accumulated in cytoplasm (a–d, immunohistochemical staining with original magnification 9400)

1934

L. Xie et al.

a

b

c Cell Count (per HPF) 150

**

Non-HGF treated HGF treated

120

* 90 60 30 CNE-1

CNE-2

FIG. 5 Invasion assay of NPC cells using Transwell chamber after treatment with exogenous HGF. a After 48 h incubation, only a few cells had passed through the 8-lm-pore membrane and attached to the undersurface in non-HGF-treated CNE-1 cells. However, in HGFtreated CNE-1 cells (b), the count of cells which had passed through the membrane was obviously increased. c Cell count evaluated in CNE-1 and CNE-2 NPC cells with or without HGF treatment. * Significantly increased count of CNE-1 cells after 48 h incubation. ** Significantly increased count of CNE-2 cells after 48 h incubation

extracellular matrix invasion have also been proven in a variety of normal and neoplastic epithelial cells in vitro.23,24 Such invasion and migration of tumor cells may be a basis for lymph node metastasis and distant metastasis. In the current study, the observed relationship between high HGF expression and lymph node metastasis as well as advanced tumor stage (stage III and IV) indicates that HGF might be a possible mediator that contributes to the extensive lymph node metastasis and accelerated progression of NPC. Moreover, on survival analysis, high HGF expression in tumor cells was closely associated with lower survival rate of patients with NPC. However, on multivariate analysis of prognostic factors in NPC, high HGF expression did not emerge as an independent factor influencing prognosis of NPC patients. These results indicate that tumor-cell-derived HGF may contribute in part to lymph node metastasis through HGF autocrine pathway and accelerate progression of NPC, but it might not be an independent prognostic factor to predict prognosis of NPC patients accurately. However, a recent study reported that immunoreactive HGF is a strong and independent predictor of recurrence and survival in human breast cancer.25

Further investigation with more cases and unabridged follow-up data needs to be carried out to clarify whether tumor species and sample size are responsible for these discrepancies. In the present study, we found that not only high HGF expression in tumor but also abnormal E-cadherin expression was significantly correlated with lymph node metastasis, advanced clinical stage, and poor prognosis of patients. These results indicate that abnormal E-cadherin expression in NPC tumor cells may also play important roles in tumor progression and influence prognosis of patients. E-cadherin and its associated intracellular catenins are regarded as maintaining cellular integrity.26 Loss of cell–cell adhesion is known to play an important role in cell invasion and dissemination from primary to metastatic sites. Downregulation of E-cadherin has been correlated with poor differentiation of tumor, advanced disease state, and unfavorable survival rate of patients in a variety of human cancers, especially carcinomas of the head and neck.27,28 Endocytosed expression of E-cadherin has recently also been revealed to be associated with tumor recurrence and poor prognosis in hypopharyngeal carcinoma, and to influence E-cadherin-mediated cell–cell adhesion in prostate cancer cells.15,18 Endocytosed E-cadherin fails to interact with complexes, which consist of anchoring proteins including a-, b-, p120-, and c-catenin, which facilitates dissociation of tumor cells from each other, promoting tumor invasion and metastasis, even though in this situation the expression of catenins does not change in tumor cells.29 Our data suggest that abnormal Ecadherin expression in tumor cells is a strong independent predictor for survival of patients on multivariate analysis of prognostic factors in NPC. Although we did not detect expression and phosphorylation of catenins in NPC tumor cells, it is reasonable to believe that only abnormal Ecadherin expression, including absent and endocytosed expression, resulting in dysfunction of E-cadherin complexes, plays important roles in metastasis and progression of NPC. E-cadherin expression status in tumor cells was valuable for prognostic evaluation of NPC patients. In this study, we found exogenous HGF to have an effect on downregulation of E-cadherin, endocytosis of E-cadherin, and promotion of NPC cell dissociation in vitro. Morphological alteration from polygonal epithelial-like to spindle mesenchymal-like was also observed in NPC cells at 48 h after HGF treatment. Consequently, enhancement of cell invasion in Transwell chamber was found in both investigated HGF-treated NPC cell lines. These results suggest that HGF may be involved in the process of epithelial–mesenchymal transition (EMT) in NPC by regulating E-cadherin-mediated cell–cell adhesion, and eventually promote cell invasion and metastasis. EMT is recognized as a crucial process during cancer progression,

Altered E-cadherin Expression by HGF in NPC

and downregulation or loss of E-cadherin is of particular importance, as it can be viewed as both a cause and effect of EMT.30 On the transcriptional level, Snail/Slug, E12/ E47, and dEF-1/SIP1 suppress E-cadherin expression through direct binding to the E-cadherin promoter and induce EMT.31–33 On the posttranslational or effector level, receptor tyrosine kinase phosphorylation and modulation of the ubiquitinase Hakai cause internalization of E-cadherin from the cell surface, resulting in either trafficking of Ecadherin for endosomal recycling or degradation.34 The accumulated data imply that HGF might modulate E-cadherin in the EMT process through both of these known pathways. One piece of supporting evidence reveals that upregulation of Snail by HGF is mediated by the MAPK/ Egr-1 signaling pathway and that both Snail and Egr-1 play critical roles in HGF-induced cell scattering, migration, and invasion.35 Other studies have demonstrated that HGF enhances tyrosine phosphorylation of b-catenin and results in dissociation of b-catenin from E-cadherin.36 Although we did not investigate the molecular mechanisms involved in HGF-induced E-cadherin alteration in NPC cells, it is possible that HGF-induced downregulation and internalization of E-cadherin in NPC tumor cells is responsible for the EMT process, promoting tumor progression and consequently influencing the prognosis of NPC patients. Interestingly, in our study, we found that high HGF expression was closely associated with high c-Met expression in tumor tissues, and that increased c-Met expression was also observed in HGF-treated NPC cells in vitro. c-Met activation is known to induce cell migration and invasion, both of which are essential for tumor progression. Previous study on NPC has also demonstrated that increased c-Met protein is associated with poor outcome of advanced NPC patients.8 Since the multiple effects of HGF are transduced via activation of c-Met, we considered that the HGF/c-Met autocrine loop in NPC plays some important roles in modulating E-cadherin and facilitating dissemination of tumor cells from primary to secondary tumor by its specific pathway. Activation of cMet can cause downregulation of cadherin-mediated cell– cell adhesion. Recently, some reports revealed that c-Met co-localization with E-cadherin on the membrane required cell–cell contact maturation and HGF-induced downregulation of cell adhesion molecules accompanied with translocation expression of c-Met to the nucleus through PI3K/Akt pathway in uveal melanoma cells.37,38 However, in our study, neither membranous nor nuclear expression pattern of c-Met was observed in NPC tissues and cells, regardless of HGF treatment. Further investigation needs to be clarified whether or not tumor species is responsible for these discrepancies. In conclusion, although the precise factors responsible for invasion and metastasis in NPC have not been

1935

identified, our study indicates that the HGF/c-Met system may contribute to cell invasion in NPC by modulating Ecadherin-mediated cell–cell adhesion through downregulation and internalization of E-cadherin. Abnormal expression of E-cadherin by HGF may play some important roles in NPC progression and is a valuable predictor for prognosis evaluation of NPC patients. From this standpoint, inhibiting HGF expression or interfering with Ecadherin internalization in primary tumor appears to be a promising strategy for developing a therapeutic approach to treat this malignant tumor. ACKNOWLEDGMENT This study was supported by a grant from the Natural Science Foundation of Guangdong Province, China (5001744). The authors especially thank Dr. Xiao-ying Tian, Hong Kong Baptist University, for her help in the preparation of the manuscript.

REFERENCES 1. Yu MC, Yuan JM. Epidemiology of nasopharyngeal carcinoma. Semin Cancer Biol. 2002;12:421–9. 2. Farias TP, Dias FL, Lima RA, Kligerman J, de Sa GM, Barbosa MM, et al. Prognostic factors and outcome for nasopharyngeal carcinoma. Arch Otolaryngol Head Neck Surg. 2003;129:794–9. 3. Boros P, Miller CM. Hepatocyte growth factor: a multifunctional cytokine. Lancet. 1995;345:293–5. 4. Liu ZX, Yu CF, Nickel C, Thomas S, Cantley LG. Hepatocyte growth factor induces ERK-dependent paxillin phosphorylation and regulates paxillin-focal adhesion kinase association. J Biol Chem. 2002;277:10452–8. 5. Zhou HY, Wong AS. Activation of p70S6K induces expression of matrix metalloproteinase 9 associated with hepatocyte growth factor-mediated invasion in human ovarian cancer cells. Endocrinology. 2006;147:2557–66. 6. Steffen M, Zuehlke I, Scherdin U. Motility factors identified in supernatants of human cholangiocarcinoma cell lines. Int J Oncol. 2001;18:1107–12. 7. Birchmeier C, Birchmeier W, Gherardi E, Vande Woude GF. Met, metastasis, motility and more. Nat Rev Mol Cell Biol. 2003; 4:915–25. 8. Qian CN, Guo X, Cao B, et al. Met protein expression level correlates with survival in patients with late-stage nasopharyngeal carcinoma. Cancer Res. 2002; 62:589-96. 9. Li Z, Lin SX, Liang HZ, He JH. Expression of hepatocyte growth factor/c-Met system in nasopharyngeal carcinoma and its biological significance. Zhonghua Bing Li Xue Za Zhi. 2005; 34:75–9. 10. Gumbinera BM. Regulation of cadherin adhesive activity. J Cell Biol. 2000;148:399–404. 11. Yagi T, Takeichi M. Cadherin superfamily genes: functions, genomic organization, and neurologic diversity. Genes Dev. 2000;14:1169–80. 12. Palacios F, Price L, Schweitzer J, Collard JG, D’Souza-Schorey C. An essential role for ARF6-regulated membrane traffic in adherens junction turnover and epithelial cell migration. EMBO J. 2001;20:4973–86. 13. Anderson MR, Harrison R, Atherfold PA, Campbell MJ, Darnton SJ, Obszynska J, et al. Met receptor signaling: a key effector in esophageal adenocarcinoma. Clin Cancer Res. 2006;12:5936–43. 14. Tannapfel A, Wittekind C, Yasui W, Yokozaki H, Tahara E. Effect of hepatocyte growth factor on the expression of E- and

1936

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

P-cadherin in gastric carcinoma cell lines. Virchows Arch. 1994;425:139–44. Miura H, Nishimura K, Tsujimura A, Matsumiya K, Matsumoto K, Nakamura T, et al. Effects of hepatocyte growth factor on E-cadherin-mediated cell-cell adhesion in DU145 prostate cancer cells. Urology. 2001;58:1064–9. Min H, Hong M, Ma J, et al. A new staging system for nasopharyngeal carcinoma in China. Int J Radiat Oncol Biol Phys. 1994;30:1037–42. Kallioniemi OP, Wagner U, Kononen J, Sauter G. Tissue microarray technology for high-throughput molecular profiling of cancer. Hum Mol Genet. 2001;10:657–62. Kim CH, Kim JH, Kahng H, Choi EC. Change of E-cadherin by hepatocyte growth factor and effect on prognosis of hypopharyngeal carcinoma. Ann Surg Oncol. 2007;14:1565–74. Li Z, Bian LJ, Li Y, Liang YJ, Liang HZ. Expression of proteaseactivated receptor-2 (PAR-2) in patients with nasopharyngeal carcinoma: correlation with clinicopathological features and prognosis. Pathol Res Pract. 2009;205:542–50. Han SU, Lee HY, Lee JH, et al. Modulation of E-cadherin by hepatocyte growth factor induces aggressiveness of gastric carcinoma. Ann Surg. 2005;242:676–83. Nakamura Y, Matsubara D, Goto A, et al. Constitutive activation of c-Met is correlated with c-Met overexpression and dependent on cell-matrix adhesion in lung adenocarcinoma cell lines. Cancer Sci. 2008;99:14–22. Maroni P, Bendinelli P, Matteucci E, Desiderio MA. HGF induces CXCR4 and CXCL12-mediated tumor invasion through Ets1 and NF-kappaB. Carcinogenesis. 2007;28:267–79. Zhou HY, Wan KF, Ip CK, Wong CK, Mak NK, Lo KW, et al. Hepatocyte growth factor enhances proteolysis and invasiveness of human nasopharyngeal cancer cells through activation of PI3K and JNK. FEBS Lett. 2008;582:3415–22. Spix JK, Chay EY, Block ER, Klarlund JK. Hepatocyte growth factor induces epithelial cell motility through transactivation of the epidermal growth factor receptor. Exp Cell Res. 2007;313: 3319–25. Yamashita J, Ogawa M, Yamashita S, Nomura K, Kuramoto M, Saishoji T, et al. Immunoreactive hepatocyte growth factor is a strong and independent predictor of recurrence and survival in human breast cancer. Cancer Res. 1994;54:1630–3. Hinck L, Nathke IS, Papkoff J, Nelson WJ. Dynamics of cadherin/catenin complex formation: novel protein interactions and pathways of complex assembly. J Cell Biol. 1994;125:1327–40.

L. Xie et al. 27. Wu H, Lotan R, Menter D, Lippman SM, Xu XC. Expression of E-cadherin is associated with squamous differentiation in squamous cell carcinomas. Anticancer Res. 2000;20:1385–90. 28. Chang HW, Chow V, Lam KY, Wei WI, Yuen A. Loss of E-cadherin expression resulting from promoter hypermethylation in oral tongue carcinoma and its prognostic significance. Cancer. 2002;94:386–92. 29. Li G, Schaider H, Satyamoorthy K, Hanakawa Y, Hashimoto K, Herlyn M. Downregulation of E-cadherin and Desmoglein 1 by autocrine hepatocyte growth factor during melanoma development. Oncogene. 2001;20:8125–35. 30. Thompson EW, Newgreen DF, Tarin D. Carcinoma invasion and metastasis: a role for epithelial-mesenchymal transition? Cancer Res. 2005;65:5991–5; discussion 5995. 31. Kondo M, Cubillo E, Tobiume K, et al. A role for Id in the regulation of TGF-beta-induced epithelial-mesenchymal transdifferentiation. Cell Death Differ. 2004;11:1092–101. 32. Horiguchi K, Shirakihara T, Nakano A, Imamura T, Miyazono K, Saitoh M. Role of Ras signaling in the induction of snail by transforming growth factor-beta. J Biol Chem. 2009;284:245–53. 33. Shirakihara T, Saitoh M, Miyazono K. Differential regulation of epithelial and mesenchymal markers by deltaEF1 proteins in epithelial mesenchymal transition induced by TGF-beta. Mol Biol Cell. 2007;18:3533–44. 34. Wells A, Yates C, Shepard CR. E-cadherin as an indicator of mesenchymal to epithelial reverting transitions during the metastatic seeding of disseminated carcinomas. Clin Exp Metastasis. 2008;25:621–8. 35. Grotegut S, von Schweinitz D, Christofori G, Lehembre F. Hepatocyte growth factor induces cell scattering through MAPK/Egr-1-mediated upregulation of Snail. EMBO J. 2006;25: 3534–45. 36. Hiscox S, Jiang WG. Hepatocyte growth factor/scatter factor disrupts epithelial tumour cell-cell adhesion: involvement of beta-catenin. Anticancer Res. 1999;19:509–17. 37. Ye M, Hu D, Tu L, et al. Involvement of PI3 K/Akt signaling pathway in hepatocyte growth factor-induced migration of uveal melanoma cells. Invest Ophthalmol Vis Sci. 2008;49:497–504. 38. Reshetnikova G, Troyanovsky S, Rimm DL. Definition of a direct extracellular interaction between Met and E-cadherin. Cell Bio Int. 2007;31:366–73.