High expression of EZH2 is associated with tumor proliferation and prognosis in human oral squamous cell carcinomas

Human OSCC cell lines 

Three human OSCC cell lines, HSC3 (poorly differentiated), HSC4 (well differentiated) and SCCKN (moderately differentiated) were cultured in Dulbecco’s modified Eagle’s medium (Nissui, Tokyo, Japan) supplemented with 10% fetal bovine serum (JRH, Lenexa, KS, USA), 100U/ml penicillin, 100μg/ml streptomycin and 292μg/ml l-glutamine.

Protein sample 

We obtained frozen tissue samples of two pairs of oral non-neoplastic tissues and well-(T1), or poorly differentiated (T2) SCCs. The frozen tissue samples and three oral SCC cell lines were solubilized in lysis buffer (50mM Tris–HCl pH 7.4, 125mM NaCl, 0.1% NP-40, 1mM PMSF, 1ng/ml leupeptin, 1ng/ml aprotinin) with a homogenizer on ice. Lysates were centrifuged at 12,000rpm for 10min and then supernatants were decanted. Protein concentration was determined using the Bradford protein assay (Bio-Rad Lab, Richmond, CA, USA) with bovine serum albumin as the standard protein.

Western blot analysis 

Fifty micrograms of protein was separated by electrophoresis on a 12% sodium dodecyl sulfate–polyacrylamide gel. The protein was then electro-transferred onto a polyvinylidene difluoride membrane (Immobilon; Millipore, Bedford, MA). After 1h incubation in blocking solution (5% non-fat dry milk in PBS, 0.5% Tween 20), the membrane was blotted with anti-beta-actin monoclonal antibody (1:1000; AC15, Sigma, St. Louis, MO) and anti-EZH2 antibody, (1:1000; clone 11, BD Biosciences, San Jose, CA). The blots were developed with peroxidase-labeled secondary antibodies. After extensive washing, specific bands were detected using an enhanced chemiluminescence system (ECL detection system, Amersham Biosciences, UK).

Tissue samples 

We analyzed tissue samples of 10 pieces of normal oral mucosa, 50 dysplasias and 102 OSCCs. All specimens were obtained from the Division of Oral and Maxillofacial Biopathological Surgery, Tottori University Faculty of Medicine. All specimens were fixed in 10% buffered formalin and embedded in paraffin wax. Paraffin blocks were sectioned in 4μm slices. All tumors were classified according to the International Union Against Cancer (UICC) tumor size nodal metastasis distant metastasis (TNM) classification.27 Histological diagnoses of OSCC and dysplasia were made according to the criteria of the World Health Organization for the histological typing of cancer and precancer of the oral mucosa.28 The histological mode of invasion was classified according to the classification of Jakobsson.29

Immunohistochemistry 

Paraffin-embedded tissue sections were immunostained using the streptavidin–biotin–peroxidase complex (SAB) technique with a mouse monoclonal antibody against EZH2 (1:25, BD Biosciences, clone 11, San Jose, CA), Ki-67 monoclonal antibody (1:50, MIB-1, Immunotech, Cedex 9, France) and P53 monoclonal antibody (1:50, DO7, Dako, Kyoto, Japan). The sections were deparaffinized, and antigens were retrieved by autoclaving in 10mM citrate buffer (pH 6.0) at 121°C for 10min. Endogenous peroxidase activity was blocked by immersing the slides in 0.3% hydrogen peroxide in methanol for 30min. The sections were reacted with each primary antibody overnight at 4°C. Tissue sections were treated with secondary antibody and biotin–streptavidin complex for 30min each at 37°C. Diaminobenzidine was used as the chromogen for the immunoperoxidase reaction. The slides were counterstained with hematoxylin.

At least 1000 normal squamous, dysplastic or SCC cells were counted, and the percentage of cells showing positive staining was designated as the labeling index (LI) (%). EZH2 immunoreactivity was classified into two groups: lower expression, when positive cells were less than 50%; and higher expression, when positive cells were 50% or more.

TUNEL analysis 

For the detection of apoptotic cells in situ, terminal deoxynucleotidyl transferase (TdT)-mediated dUTP-digoxigenin nick-end labeling (TUNEL) was performed using an Apop Tag Plus in situ apoptosis detection kit (Chemicon International, Temecula, CA). Briefly, after deparaffinization, incubation was carried out with 20μg/ml proteinase K (Wako Pure Chemical Industries, Osaka, Japan) for 10min at 37°C. Then, the sections were incubated with terminal deoxynucleotidyl transferase, digoxigenin-11-dUTP and dATP in a moist chamber for 60min at 37°C. Incubation with anti-digoxigenin antibody peroxidase for 30min at room temperature was employed for the detection of digoxigenin-11-dUTP labeling, followed by color development with a solution containing 3,3′-diaminobenzidine and H2O2. Hematoxylin was used for counterstaining.

At least 1000 normal, dysplastic or SCC cells were counted, and the percentage of cells showing positive staining was designated as the apoptotic index (AI) (%).

Statistical analysis 

Statistical analyses were performed using Mann–Whitney’s U test and the Kruskal–Wallis test. The probability of survival by EZH2 expression was determined using the Kaplan–Meier method. Different survival curves were compared using the log-rank test. A difference of p<0.05 was considered statistically significant.

Expression of EZH2 in oral SCC 

Western blot analysis detected EZH2 protein as a single band with a molecular weight of 91kDa. As shown in Figure 1, EZH2 protein was found in HSC3 (well-differentiated SCC), HSC4 (poorly differentiated SCC), SCCKN (moderately differentiated SCC) cell lines and both resected SCC tissue specimens. In contrast, the expression of EZH2 protein was almost absent in non-neoplastic mucosae compared to the corresponding OSCCs.

View Large Image Download to PowerPoint

Figure 1. EZH2 expression in three SCC cell lines, non-neoplastic cells (N1, N2) and well-(T1) and poorly differentiated (T2) SCCs, in two pairs of human oral mucosae, and in SCCs was examined by Western blotting. EZH2 protein expression was found in the three OSCC cell lines and two surgical specimens. By contrast, the expression of EZH2 was almost absent in non-neoplastic mucosae (N1 and N2).

Expression of EZH2 in normal epithelia, dysplasias and SCCs 

Immunohistochemistry was performed to examine the expression of EZH2 protein and its distribution in human oral mucosae and OSCC. Immunoreactivity for EZH2 was observed in the nuclei of normal epithelial cells and dysplastic cells and SCC cells (Fig. 2). EZH2-positive cells were distributed in basal and suprabasal cell layers in normal mucosae (Fig. 2a) and in both basal cell and prickle cell layers in dysplasia (Fig. 2b). EZH2-positive cells extended from the lower layer to the upper layer, correlating with the grade of dysplasia. EZH2-positive cells were observed mainly in peripheral portions of cancer nests in well- and moderately-differentiated OSCCs (Fig. 2d and g). In contrast, no specific distribution of positive cells was noted in poorly differentiated OSCC (Fig. 2j). Interestingly, the frequency of EZH2-positive cells was higher than Ki-67-positive cells in each differentiation of OSCC (Fig. 2).

View Large Image Download to PowerPoint

Figure 2. Immunohistochemistry of EZH2 in human oral normal mucosa (a), dysplasia (b) and SCC (c–k). EZH2-positive cells presented mainly in the basal and suprabasal cell layers in normal mucosa (a), in contrast to presenting from the basal cell to prickle layers in dysplasia (b). EZH2 immunoreactivity was shown in the nuclei of both normal epithelia (a) and dysplasia (b). EZH2 immunoreactivity was noted mainly in the nuclei of SCC cells (d,g,j). EZH2-positive cells were present in the marginal areas of cancer nests in well-differentiated SCC (d). The number of EZH2-positive cells was higher in moderately differentiated SCCs than in well-differentiated SCC (g). In poorly differentiated SCC, no specific localization of positive cells was observed and most tumor cells showed positive immunoreactivity (j). The immunoreactivity of EZH2 and Ki-67 in carcinoma was almost similar, and the number of EZH2-positive cells was higher than Ki-67-positive cells. HE: (c), (f) and (i). EZH2 immunostaining: (d), (g) and (j). Ki-67 immunostaining: (e), (h) and (k). Well-differentiated SCC: (c–e), moderately differentiated SCC: (f–h), poorly differentiated SCC; (i–k).

Table 1 shows the LIs of EZH2, Ki-67 and P53 and the AI in oral normal mucosae, dysplasias and SCCs. The mean LI of EZH2 was 19.4±1.4, 25.9±3.0, and 50.7±2.1 in normal mucosae, dysplasias and OSCCs, respectively. The LI of EZH2 was significantly higher in OSCCs than in normal mucosae and dysplasias, respectively (p<0.05), but no significant difference was noted between normal mucosae and dysplasias. The LI of EZH2 was highest in OSCC, followed by dysplasias and normal epithelia. EZH2 LIs also correlated with the histological grade of dysplasia and OSCC, being highest for severe dysplasia and poorly differentiated OSCC. Similar results were also obtained for LIs of Ki-67 and P53, but the value of EZH2 LI was highest among those factors. On the other hand, no significant difference was noted between AIs among normal mucosa, dysplasia and OSCC.

Table 1. LI of EZH2, AI and LI of Ki-67, and P53 in oral normal epithelia, dysplasias and SCCs

Mean±SE%, ap<0.01 and bp<0.05 are statistically significant in Mann–Whitney’s U test. LI: labeling index and AI: apoptotic index.

Correlation of EZH2 expression levels with clinicopathological parameters in OSCCs 

Immunohistochemical results and their associations with clinicopathological features in 102 OSCCs are summarized in Table 2. The cut-off level of EZH2 LIs was fixed at 50%, which was almost the median (49.6%) for all cases. Fifty-one cases had a higher expression of EZH2, which significantly correlated with histological differentiation (p=0.002), clinical stage (p=0.018), tumor size (p=0.038) and lymph node metastasis (p=0.041), but not with age, sex and mode of invasion.

Table 2. The relationship between EZH2 expression and clinicopathological data in 102 oral SCCs

	Characteristics	Cases	EZH2		p-valuea
			Higher (≧50%)	Lower (<50%)
	Age (years)
	≧65	53	30	23	0.167
	<65	49	21	28
	Sex
	Male	67	38	29	0.062
	Female	35	13	22
	Histological differentiation
	Well	45	14	31	0.002
	Moderate	34	20	14
	Poor	23	17	6
	Clinical stage
	I	19	5	14	0.018
	II	28	11	17
	III	17	12	5
	IV	38	23	15
	Tumor size
	T1	21	6	15	0.038
	T2	40	17	23
	T3	11	9	2
	T4	30	19	11
	Lymph node involvement
	N0	64	27	37	0.041
	N1, N2 and N3	38	24	14
	Mode of invasion
	Grades 1 and 2	10	3	7	0.221
	Grade 3	71	34	37
	Grade 4	21	14	7

Higher: positive cells are 50% or more and lower: positive cells are less than 50%.

a p<0.05 was regarded as statistically significant in Kruskal–Wallis test.

Correlation among LIs of EZH2, Ki-67, P53 and AI in oral SCC 

Table 3 shows the correlation between EZH2 expression and the LIs of Ki-67 and P53, and AI in 102 OSCCs. The mean LIs of Ki-67 and P53 were 35.1±2.1 and 29.3±2.9 in the 51 SCCs with higher EZH2 expression, and 25.1±1.8 and 19.9±2.2 in the 51 SCCs with lower EZH2 expression. There was a positive correlation between EZH2 expression and the LIs of Ki-67 and P53 in OSCC, respectively (p<0.01). Table 3 shows that the Spearman’s correlation coefficient between EZH2 and Ki-67 was rs=0.38 (p<0.01) and P53 was rs=0.29 (p<0.01). There was a statistically significant difference between LIs of EZH2 and Ki-67 or P53. In contrast, no significant correlation was noted between LIs of EZH2 and AI.

Table 3. Correlation between EZH2 expression and LIs of Ki-67/P53 and AI in oral SCC

		EZH2		p-valuea	Spearman correlation
		Higher (N=51) (mean±SE)	Lower (N=51) (mean±SE)		rs	p-valueb
	Ki-67LI	35.1±2.1	25.1±1.8	<0.01	0.38	<0.01
	P53LI	29.3±2.9	19.9±2.2	<0.01	0.29	<0.01
	AI	0.48±0.4	0.46±0.1	0.95	0.03	0.75

Median: mean±SE% and rs: Spearman rank-correlation coefficient.

a p<0.01 is statistically significant in Mann–Whitney’s U test. b p<0.01 is statistically significant in Spearman rank-correlation.

Survival analysis 

We performed a prognostic analysis in 83 patients with stages II–IV OSCCs. We excluded the 19 stage I cases, because of their favorable prognosis. Kaplan–Meier survival curves showed the 100-months survival rate analysis of OSCCs. We examined the correlation between the expression of EZH2, Ki-67, P53 and the AI with cumulative survival. The median value (EZH2=50%, P53=19.5%, Ki-67=28.2%, AI=0.43%) was set as the cut-off value. As shown in Figure 3a, higher LI of EZH2 showed a significantly poorer prognosis in cumulative survival time. On the other hand, there were no significant differences in the Kaplan–Meier survival curves of Ki-67 and P53 LIs, although prognosis tended to be poor with higher LIs of Ki-67 and P53 (Fig. 3b and c). The value of AI did not correlate with the cumulative survival time in OSCC patients (Fig. 3d).

View Large Image Download to PowerPoint

Figure 3. Kaplan–Meier survival curves of the 83 stage II–IV oral SCCs. Patients with a higher expression of EZH2 showed significantly unfavorable prognoses (a). On the other hand, no significant differentiation was observed in the expression of Ki-67 (b), P53 (c), and the number of AI (d) in cumulative survival time.