Estrogen receptor α and β in esophageal squamous cell carcinoma

Abstract

A gender difference has been reported in the morbidity of esophageal squamous cell carcinoma (ESCC). Estrogens have been proposed to play a role in this difference but the details have not yet been clarified. Therefore, in the present study, we examined the status of estrogen receptor (ER)α and ERβ in 90 Japanese ESCC patients. ERα and ERβ immunoreactivity was detected in the nuclei of ESCC cells (41.1 and 97.8%, respectively). There was a significant positive association between the ERβ H score and histological differentiation (P = 0.0403), TNM‐pM (LYM) (P = 0.00164) and Ki67/MIB1 LI of carcinoma cells (P = 0.0497, r = 0.207). In addition, the ERβ status of carcinoma cells was significantly correlated with unfavorable clinical outcome of the patients. Multivariate analysis further revealed the ERβ status in carcinoma cells as an independent unfavorable prognostic factor of these patients. We further examined the effects of estrogen treatment on ESCC cell line (ECGI‐10) transfected with ERα or ERβ in vitro. The number of ECGI‐10 transfected with ERβ was increased by estradiol or ERβ specific agonist but estradiol did not exert any effect upon the cell number of ECGI‐10 transfected with ERα. In summary, the results of the present study clearly demonstrate that the status of ERβ in ESCC was closely associated with the unfavorable prognosis, possibly through altering cell proliferation of carcinoma cells. (Cancer Sci 2012; 103: 1348–1355)

Human esophageal squamous cell carcinoma (ESCC) is one of the most aggressive malignancies, despite recent marked improvement of therapeutic techniques and perioperative management. Results of most epidemiological studies demonstrate prominent gender differences in the prevalence of ESCC. ESCC is generally more common in men than women, with a male to female ratio of approximately 6:1 in Japan.1 In addition, the prognosis of ESCC patients is reported to be significantly better in women than in men following an adjustment of various clinicopathological factors.2 Several reported studies also suggest the possible roles of estrogens in development of ESCC.3, 4

Sex steroids, such as estrogen, are well‐known to contribute to physiological maturation and cell proliferation of estrogen dependent tissues, such as breast and endometrial tissues. Estrogens also play important roles in many non‐gonadal or non‐classical estrogen' receptors' dependent tissues in both men and women through estrogen receptors (ER). Therefore, it is important to examine the status of ER in these tissues. Two different isomers of ER, ERα and β, were first identified by Enmark et al.5 Both ERα and ERβ are encoded by ESR1 located on chromosome 6q256 and ESR2 located on chromosome 14q22‐24,7 respectively. Estrogens have also been reported to play pivotal roles in several types of human malignancies, which had not been necessarily considered as classical estrogen‐dependent neoplasms, such as those arising in lung,8, 9 urinary bladder10 and gastrointestinal tract.11, 12 ERwere also reported to be detected in laryngeal and ESCC.13, 14, 15, 16, 17 Nozoe et al. report that the group of the patients associated with cytoplasmic ERα‐positive/nuclear ERβ‐negative status was associated with a significant adverse clinical outcome in 73 cases of ESCC.13 Kalayarasan et al. also report that the status of ERβ in carcinoma cells was correlated with poor histological differentiation and advanced clinical stages in ESCC patients.15 These reported findings all suggest that estrogens are considered to be at least involved in biological behavior of ESCC through ER present in carcinoma cells. However, it is also true that controversy exists regarding the biological and clinical significance of estrogens in ESCC.17

Therefore, in the present study, we attempted to evaluate estrogen actions through ERα and ERβ in ESCC as follows. We first examined the status of ERα and ERβ using immunohistochemistry in 90 Japanese ESCC cases and evaluated their correlation with clinicopathological findings of individual patients, including overall survival or disease‐free survival. We further characterized the potential biological functions of ERα and ERβ in ESCC cell lines stably transfected with ERα or ERβ.

Materials and Methods

Patients and tissue preparation

A total of 90 specimens of thoracic ESCC were obtained from Japanese patients who underwent potentially curative esophagectomy with lymph node dissection from 2000 to 2005 at the Second Department of Surgery at Tohoku University Hospital (Sendai, Japan). These patients had received neither chemotherapy nor irradiation therapy prior to surgery. These 77 men and 13 women had a median age of 63.9 years (range, 37–81 years). A total of 18 specimens of non‐neoplastic epithelium were also obtained from these 90 cases. These specimens had been fixed with 10% formalin for 36–48 h at room temperature and embedded in paraffin wax. Relevant clinical data were retrieved from careful review of the patients' charts. Histopathological features of all the tumors were independently reviewed by four of the authors (M. Z., D. T., F. F. and H. S.). The pathological stage of each cancer was defined according to the TNM system and each lesion was graded histologically according to the World Health Organization classification. The median follow‐up time was 65.5 months (range, 1–119 months). The research protocol in this study was approved by the Ethics Committee at the Tohoku University School of Medicine (Accesssion No. 2009‐453) and informed consent was obtained from each patient before surgery.

Immunohistochemistry

Mouse monoclonal antibodies for ERα(6F11), ERβ(14C8) and Ki‐67 (MIB1) were purchased from Novocastra (Newcastle, UK), Gene Tex (San Antonio, TX, USA) and DAKO (Carpinteria,CA, USA), respectively.

Serial 4‐μm thick tissue sections were deparaffinized with xylene and ethanol. Antigen retrieval was performed by heating the slides in an autoclave at 121°C for 5 min in citric acid buffer (2 mmol/L citric acid and 9 mmol/L trisodium citrate dehydrate, pH 6.0) or instant antigen retrieval H buffer (Mitsubishi Kagaku Iatron, Tokyo, Japan) for ER. Sections were then incubated with 10% normal rabbit serum for the monoclonal antibodies to reduce nonspecific background immunostaining. All incubations were performed for 18 h at 4°C with primary antibodies. The dilutions of primary antibodies were summarized as follows: ERα, 1:50; ERβ, 1:1000; and Ki‐67, 1:100. Endogenous peroxidase activity was blocked by immersing the slides in 0.3% hydrogenperoxidase for 30 min at room temperature. The sections were subsequently incubated with biotinylated rabbit antimouse IgG (Histofine Kit; Nichirei, Tokyo, Japan). The localization of ER was then visualized with 3‐Amino‐9‐ethylcarbazole and counterstained with hematoxylin. The other antigen‐antibody complex was then visualized with 3.3‐diaminobenzidine (1 mmol/L, in 50 mol/L Tris–HCl buffer, pH 7.6 and 0.006% H₂O₂) and counterstained with hematoxylin. ERα positive breast cancer tissue was used as a positive control for ERα. Non‐pathologic prostate tissue was used as a positive control for ERβ. As a negative control, normal mouse or rabbit IgG was used instead of the primary antibodies and no specific immunoreactivity was detected in these preparations.

Evaluation of immunoreactivity

We evaluated the nuclear immureactivity of ERα, ERβ and Ki67. The immunoreactivities for ERα and Ki67 were evaluated in approximately 1000 carcinoma cells in each case and the percentage of immunoreactivity (i.e. labeling index [LI]) was determined after counting. In the present study, we used the H score for semi‐quantitative analysis of nuclear ERβ immunoreactivity.18 Approximately 1000 tumor cells were counted in three different representative areas of carcinoma infiltration in each case and the H score was derived from the formula: H score = (percentage of strongly stained nuclei × 3) + (percentage of moderately stained nuclei × 2) + (percentage of weakly stained nuclei × 1). Scoring was performed independently by three of the authors (M. Z., D. T. and F. F.). When the inter‐observer differences were <5%, the mean value was obtained as a final value of LI and H score. When the differences were more than 5%, all of the examiners evaluated the immunohistochemical findings simultaneously through multi‐headed light microscopy and reached a consensus. When evaluating the possible correlation between ERβ status and clinical outcome of individual patients, the cases were tentatively classified into two groups according to their ERβ H scores (high ERβ, ≥250 H score; low ERβ, 0–249 H score, respectively), because ERβ H scores were found to be distributed in a peak on the boundary of approximately 250 (data not shown). In addition, the median value of the ERβ H score was 231.2, near the value of 250. The selection of the median value as the cut‐off point was reported to secure objectivity because it is not determined as an “optimal” cut‐off point.19, 20

Cell culture and chemicals

Human ESCC cell line EC‐GI‐10 was provided by RIKEN Bioresource Center (Tsukuba, Japan). TE‐1, TE‐4 and TE‐8 were obtained from the Cell Resource Center for Biomedical Research, Institute of Development, Aging and Cancer, Tohoku University (Sendai, Japan). All ESCC cell lines were cultured in RPMI 1640 (Sigma‐Aldrich, St. Louis, MO, USA) with 10% FBS (Nichirei, Tokyo, Japan). In this study, cells were cultured with phenol red‐free RPMI 1640 containing 10% dextran coated charcoal treated FBS for 3 days before the experiment. 17β‐estradiol was purchased from Sigma‐Aldrich. An ERα selective agonist (propyl‐pyrazole‐triol; PPT),21 an ERβ selective agonist (diarylpropionitrile [DPN])21 and a pure ER antagonist, ICI 182780,22 were purchased from Tocris Bioscience (Minneapolis, MN, USA).

Stable transfection: establishment of EC‐GI‐10 cells expressing ERα or ERβ

The transformed ECGI‐10 cells expressing ERα (EC‐GI‐10 + ERα) or ERβ (EC‐GI‐10 + ERβ) were established to further characterize the biological functions of ER isoforms in ESCC.

Stable transfection was performed according to the methods reported in previous studies with some modifications.8, 23, 24 ERα and ERβ expression vectors for ERα (pRc/CMV‐ERα) and ERβ (pRc/CMV‐ERβ) used in this study were described previously.8, 23, 24 Briefly, EC‐GI‐10 cells were transfected with ERα or ERβ expression vector with Lipofectamine LTX (Invitrogen Life Technologies, Gaithersburg, MD, USA). After 24 h in culture, the cells were subsequently grown in fresh RPMI 1640 supplemented with 10% FBS containing 1 mg/mL geneticin (G418; Sigma‐Aldrich) for 2 weeks. Isolated colonies were trypsinized in metal ring cups and the cells were further cultured in the presence of 200 μg/mL G418. As a negative control, empty vector was also transfected in the EC‐GI‐10 cells. Expression of ERα and ERβ at both mRNA and protein levels in EC‐GI‐10 cells was examined by quantitative RT‐PCR and immunohistochemistry.

RT‐PCR

Total RNA was carefully extracted from human ESCC cell line (EC‐GI‐10, TE‐1, TE‐4 and TE‐8) using the TRIzol (Invitrogen Life Technologies) method. The QuantiTect Reverse Transcription kit (Qiagen, Hilden, Germany) was used in the synthesis of cDNA. RT‐PCR was carried using the Light‐Cycler System (Roche Diagnostics, Manheim, Germany), and ribosomal protein L 13a (RPL13A) was also used as an internal standard. The primer sequences used in this study are as follows: ERα(NM_000125); forward: 5′‐AGACACTTTGATCCACCTGA‐3′ (cDNA position 1811–1831) and reverse: 5′‐CAAGGAATGCGATGAAGTAG‐3′ (cDNA position 2080–2100), ERβ (AB006590); forward: 5′‐CCTGGCTAACCTCCTGATGC‐3′ (cDNA position 1460–1480) and reverse: 5′‐ACCCCGTGATGGAGGACTT‐3′ (cDNA position 1608–1627) and RPL13A (NM_012423); forward: 5′‐CCTGGAGAGAAGAGGAAAGAGA‐3′ (cDNA position 487–509) and reverse: 5′‐TTGAGGACCTCTGTGTATTTGTCAA‐3′ (cDNA position 588–612). The PCR products were purified and subjected to direct sequencing to verify amplification of the correct sequences. Negative controls, in which the reaction mixture lacked cDNA template, were included to exclude the possibility of exogenous contaminant DNA.

cDNA of known concentrations for target genes and the housekeeping gene RPL13A were used to generate standard curves for real‐time quantitative PCR to determine the quantity of target cDNA transcript. The Ct (cycle threshold) values were used to calculate the gene‐specific input mRNA amounts according to the calibration curve method. The mRNA level in each case was represented as a ratio of RLP13A and was evaluated as a ratio (%) compared with that of each control.8, 20, 25

Cell proliferation assays

EC‐GI‐10 including its transformants and TE‐1, TE‐4 and TE‐8 were treated with the indicated compounds for 3 days and the cell proliferation was measured by a WST‐8 [2‐(2‐methoxy‐4‐nitrophenyl)‐3‐(4‐nitrophenyl)‐5‐(2,4‐disulfophenyl)‐2H‐tetrazolium, monosodium salt] method using Cell Counting Kit‐8 (Dojin Kagaku, Kumamoto, Japan).8, 26

Statistical analysis

The values for ERβ H score, patient age, tumor size, Ki67/MIB1 LI and ERα LI were demonstrated as mean ± SEM. The statistical analyses between ERβ H score and clinicopathological parameters of individual patients were evaluated using the Mann–Whitney U‐test, the Kruskal–Wallis test, a correlation coefficient (r) and regression equations when appropriate.

Overall survival (OS) and disease‐free survival (DFS) curves of the patients examined were generated according to the Kaplan–Meier method and statistical significance was calculated using the log‐rank test. Cox's proportional hazard model was used for both univariate and multivariate analyses.

The proliferation activity was evaluated as a ratio (%) compared to that of controls (no treatment with either estradiol for 72 h). The statistical analyses between relative luciferfase activity and each agent were evaluated using the Kruskal–Wallis test and the Scheffe test.

Statistical differences were examined using statview 5.0 J software (SAS Institute, Cary, NC, USA) and values of P < 0.05 were considered statistically significant.

Results

Immunohistochemistry: ERα and ERβ status in non‐neoplastic esophagus and esophageal squamous cell carcinoma

Both ERα and ERβ immunoreactivity were detected in the nuclei of non‐neoplastic basal layer cells of the esophagus (Fig. 1a,b). ERα immunoreactivity was detected in the nuclei of carcinoma cells in 38/90 ESCC (Fig. 1c). The mean value of the ERα labeling index in 90 ESCC was 22.3 ± 3.2 (range, 0–90). ERβ immunoreactivity was detected in the nuclei of carcinoma cells with a variety of immunointensity in 88/90 ESCC (Fig. 1d). The mean value of the ERβ H score in 90 ESCC was 208.9 ± 7.4 (range, 0–295).

Figure 1.

Representative illustrations of immunohistochemistry in human esophageal squamous cell carcinoma. (a) Estorogen receptor α (ERα) immunoreactivity was detected in the nuclei of non‐neoplastic epithelial cells. (b) Estorogen receptor β (ERβ) immunoreactivity in the nuclei of non‐neoplastic epithelial cells. (c) ERα immunoreactivity in esophageal squamous cell carcinoma (ESCC) case. (d) ERβ immunoreactivity in ESCC case. Black arrows represent marked immunoreactivity (+++) of ERβ and white arrows represent weak immunoreactivity (+) of ERβ. Bar represents 100 μm.

Correlation between the status of ERα immunoreactivity and clinicopathological variables in 90 esophageal squamous cell carcinoma patients

Associations between ERα status and clinicopathological variables of the patients examined are summarized in Table 1. No significant association was detected between ERα status and age, gender, tumor size, depth of tumor invasion, presence of lymph node metastasis, TNM stage, lymphatic invasion, venous invasion, infiltrative growth pattern and the Ki67/MIB1 Labeling index status of the patients examined.

Table 1.

Correlation between ERα and ERβ and clinicopathological variables in 90 esophageal squamous cell carcinoma patients

Variable	n or mean ± SEM	Nuclear ERα		Nuclear ERβ
		Labeling		Index H score
		Mean ± SEM	P‐values	Mean ± SEM	P‐values
Age (years)
<65	48	22.2 ± 4.5	0.5991	201.0 ± 10.3	0.1929
≥65	42	22.4 ± 4.7		218.3 ± 10.5
Gender
Male	77	22.6 ± 3.5	0.6298	209.2 ± 7.9	0.7786
Female	13	20.5 ± 9.0		206.8 ± 21.8
Tumor size (mm)
<50	54	25.9 ± 4.5	0.2439	208.3 ± 10.2	0.6329
≥50	36	17.0 ± 4.4		209.7 ± 10.6
TNM‐pT
pT1	29	28.7 ± 6.5	0.4309	202.2 ± 16.5	0.2148
pT2	11	25.8 ± 9.3		206.7 ± 15.4
pT3	47	19.0 ± 4.1		209.8 ± 9.1
pT4	3	0.0		266.2 ± 13.3
TNM‐pN
pN0	36	27.0 ± 5.5	0.3130	200.3 ± 13.1	0.6565
pN1	54	19.2 ± 3.9		214.6 ± 8.7
TNM‐pM
pM0	81	23.8 ± 3.5	0.3569	203.7 ± 7.9	0.0164**
pM1 (LYM)	9	8.5 ± 6.8		255.4 ± 10.4
TNM‐pStage
I	19	36.2 ± 8.4	0.2381	198.2 ± 20.8	0.0580
II	31	7.6 ± 5.1		195.5 ± 12.5
III	31	22.5 ± 5.3		215.3 ± 10.8
IV	9	8.5 ± 6.8		255.4 ± 10.4
Differentiation
Well	17	11.5 ± 4.9	0.0688	188.2 ± 23.4	0.0403**
Mod	59	20.5 ± 4.0		205.2 ± 8.5
Poorly	14	43.0 ± 9.1		249.4 ± 7.8
Lymphatic invasion
Negative	35	24.2 ± 5.8	0.8167	218.3 ± 11.3	0.1288
Positive	55	21.1 ± 3.8		202.9 ± 9.7
Venous invasion
Negative	31	26.2 ± 5.5	0.4223	209.7 ± 13.6	0.5810
Positive	59	20.3 ± 4.0		208.4 ± 8.8
Growth pattern
Expansive	26	21.3 ± 6.1	0.4961	224.4 ± 12.6	0.1039
Intermediate	52	20.4 ± 4.2		198 ± 10.0
Infiltrative	12	32.9 ± 8.9		222.4 ± 20.4
Ki‐67 LI (%)	43.7 ± 2.0 (8.1–83.9)	0.8412 (r = 0.021)		0.0497** (r = 0.207)

Data are presented as mean ± SEM. All other values represent the number of cases. ** P‐values less than 0.05 were considered significant. ER, estrogen receptor; LI, labeling index.

Correlation between the status of ERβ immunoreactivity and clinicopathological variables in 90 esophageal squamous cell carcinoma patients

Associations between ERβ status and clinicopathological variables of the individual patients examined are summarized in Table 1. There was a statistically significant positive association between ERβ H score and tumor differentiation (P = 0.0403) and TNM‐pM (LYM) (P = 0.0164). There was also a weak but statistically significant positive correlation between the ERβ H score and Ki67/MIB1 LI (P = 0.0497, r = 0.207). No significant association was detected between ERβ immunoreactivity and age, gender, tumor size, depth of tumor invasion, presence of lymph node metastasis, TNM stage, lymphatic invasion, venous invasion or infiltrative growth pattern of the patients examined in the present study.

Correlation between ERα status and clinical outcome in 90 esophageal squamous cell carcinoma patients

The OS and DFS curves of the patients examined are summarized in Figure 2(a, b), respectively. The patients with positive nuclear ERα immunoreactivity in carcinoma cells were by no means associated with better survival or favorable clinical outcome (log‐rank test: OS, P = 0.4660; DFS, P = 0.3468).

Figure 2.

Overall and disease‐free survival (OS and DFS) curves of 90 Japanese patients with esophageal squamous cell carcinoma examined in this study according to the status of estrogen receptor (ER)α and ERβ immunoreactivity (Kaplan–Meier method). Significant difference in survival of the patients was noted between those with high and low ERβ nuclear immunoreactivity (c,d), but significant difference was not detected according to the status of ERα nuclear immunoreactivity (a,b). Cases were tentatively classified into two groups according to results of ERβ H score in the nuclei: high ERβ, >250 H score and low ERβ, <250 H score. High ERβ status in carcinoma cells was significantly associated with unfavorable clinical outcome of the patients examined.

Correlation between ERβ status and clinical outcome in 90 esophageal squamous cell carcinoma patients

The OS and DFS curves of the patients examined are illustrated in Figure 2(c, d), respectively. In the present study, the patients with high nuclear ERβ immunoreactivity were significantly associated with shorter survival or adverse clinical outcome (log‐rank test: OS, P = 0.0017; DFS, P = 0.0005). Results of univariate analysis (Table 2) demonstrated that pathological stage (OS, P = 0.0003; DFS, P = 0.0006), ERβ status in the nucleus of carcinoma cells (OS, P = 0.0025; DFS, P = 0.0010), tumor size (OS, P = 0.0485; DFS, P = 0.0366) and infiltration type (OS, P = 0.0200; DFS, P = 0.0416) were all significant prognostic factors for OS and/or DFS in 90 ESCC examined in our study. A subsequent multivariate analysis did reveal that ERβ status (OS, P = 0.0010; DFS, P = 0.0007) was an independent prognostic factor for OS and DFS of these patients, as well as pathological stage (OS, P = 0.0019; DFS, P = 0.0091) and infiltration type (OS, P = 0.0185; DFS, P = 0.0328).

Table 2.

Univariate and multivariate analysis of overall survival/disease free survival in 90 esophageal squamous cell carcinoma patients

	Univariate		Multivariate
	P‐value	Relative risk (95% CI)	P‐value	Relative risk (95% CI)
(A) Overall survival
Sex (male/female)	0.1337	2.196 (0.785–6.139)
Age (≥65/<64 years)	0.9419	1.022(0.572–1.824)
Depth of tumor (pT3,4/pT1,2)	0.0069**	2.418 (1.274–4.589)
TNM–pN (positive/negative)	0.0105**	2.287 (1.213–4.311)
TNM–pM (positive/negative)	0.0579	2.191 (0.974–4.927)
pStage (III,IV/I,II)	0.0003**	3.065 (1.675–5.609)	0.0019**	2.786 (1.459–5.322)
Tumor size (≥50/<50 mm)	0.0485**	1.807 (1.004–3.252)	0.4515	1.279 (0.674–2.427)
Lymphatic invasion (positive/negative)	0.0521	1.866 (0.994–3.503)
Venous invasion (positive/negative)	0.3368	1.361 (0.726–2.552)
Growth pattern (infiltrative/expansive + intermediate)	0.0200**	2.428 (1.150–5.126)	0.0185**	2.531 (1.169–5.484)
Histological grade (por/well + mod)	0.3653	1.424 (0.662–3.062)
Ki67 status†	0.7021	0.997 (0.484–1.540)
ERα status (positive/negative)	0.4692	1.249 (0.684–2.281)
ERβ status (high/low)	0.0025**	2.469 (1.374–4.438)	0.0010**	2.754 (1.509–5.027)
Brinkman index (≥400/<400)	0.9389	0.975 (0.504–1.885)
(B) Disease free survival
Sex (male/female)	0.1142	2.286 (0.819–6.378)
Age (≥65/<64 years)	0.9997	1.000 (0.567–1.763)
Depth of tumor (pT3,4/pT1,2)	0.0124**	2.200 (1.186–4.080)
TNM–pN (positive/negative)	0.0218**	2.051 (1.110–3.788)
TNM–pM (positive/negative)	0.1305	1.865 (0.831–4.185)
pStage (III,IV/I,II)	0.0006**	2.804 (1.560–5.039)	0.0091**	2.304 (1.230–4.314)
Tumor size (≥50/<50 mm)	0.0366**	1.840 (1.038–3.259)	0.3092	1.378 (0.743–2.555)
Lymphatic invasion (positive/negative)	0.1816	1.506 (0.826–2.746)
Venous invasion (positive/negative)	0.4967	1.236 (0.671–2.276)
Growth pattern (infiltrative/expansive + intermediate)	0.0416**	2.150 (1.029–4.490)	0.0328**	2.307 (1.071–4.971)
Histological grade (por/well + mod)	0.1013	1.844 (0.887–3.835)
Ki67LI†	0.6212	0.996 (0.981–1.012)
ERα status (positive/negative)	0.3563	1.317 (0.734–2.363)
ERβ status (high/low)	0.0010**	2.627 (0.734–2.363)	0.0007**	2.828 (1.551–5.158)
Brinkman index (≥400/<400)	0.8198	1.079(0.560–2.078)

**Data considered significant (P < 0.05) in the univariate analysis were examined in the multivariate analysis. CI, confidence interval; ER, estrogen receptor; LI, labeling index; mod, moderate; por, poor. ERα status (positive; LI ≥ 10%/negative; LI < 10%). ERβ status (high; H score ≥ 250/low; H score < 250). †Data were evaluated as continuous values and displayed mean ± SEM (range).

Establishment of EC‐GI‐10 esophageal squamous cell carcinoma cells expressing ERα or ERβ

The transformed EC‐GI‐10 ESCC cells expressing ERα (EC‐GI‐10 + ERα) or ERβ (EC‐GI‐10 + ERβ) were established because the parental EC‐GI‐10 cells and TE‐1, TE‐4 and TE‐8 cells examined in this study expressed only miniscule amounts of ERα or ERβ at mRNA levels (data not shown). Estradiol administration (1 pmol/L–100 nmol/L) did not change the number of cells even after 72 h in all the cell lines, TE‐1, TE‐4, TE‐8 and EC‐GI‐10 (data not shown). As a control, we also isolated a clone tentatively termed EC‐GI‐10 + CMV, which was stably transfected with empty vector in the EC‐GI‐10 cells.

Relatively low levels of ERα and ERβ mRNA were detected in EC‐GI‐10 + CMV cells (Fig. 3A). The levels of ERα and ERβ mRNA were significantly higher in the EC‐GI‐10 + ERα and + ERβ cells, respectively, than those in EC‐GI‐10 + CMV cells (Fig. 3A). In immunohistochemistry, ERα and ERβ immunoreactivity was relatively weak or negative in EC‐GI‐10 + CMV cells (Fig. 3B). Immunoreactivity of ERα but not of ERβ was detected in EC‐GI‐10 + ERα cells (Fig. 3B) and vice versa in EC‐GI‐10 + ERβ cells (Fig. 3B).

Figure 3.

(A) mRNA level of EC‐GI‐10 + CMV versus EC‐GI‐10 + ERα(left), mRNA level of EC‐GI‐10 + CMV versus EC‐GI‐10 + ERβ(right), mean ± SD (n = 6). (B) Immunostaining for estrogen receptor (ER) isoforms in the EC‐GI‐10 transformants. (a,d) EC‐GI‐10 + CMV; (b,e) EC‐GI‐10 + ERα; (c,f) EC‐GI‐10 + ERβ. ERα immunoreactivity was detected in the nuclei of EC‐GI‐10 + ERα cells (b), ERβ immunoreactivity was detected in the nuclei of EC‐GI‐10 + ERβ cells (f). No significant immunoreactivity for ER isoforms was detected in EC‐GI‐10 + CMV cells (a,d). Immunoreactivity was evaluated in the cell blocks specimens. Bar represents 100 μm. (C) Left, EC‐GI‐10 + ERα cells; center, EC‐GI‐10 + ERβ cells; right, EC‐GI‐10 + CMV. Estradiol (1 μmol/L) with or without ER antagonist ICI 182780 (100 pmol/L), ERα agonist PPT (1 μmol/L), ERβ agonist DPN (1 μmol/L) were added to these cells. They were then cultured for 72 h. NC, no changes. The cell proliferation activity was evaluated as a ratio (%) compared with that of controls (no treatment with either estradiol, PPT, DPN or ICI 182780 for 72 h). Mean ± SD (n = 6). *P < 0.05 versus controls (Kruskal–Wallis test and Scheffe test). (D) Effects of the estrogen‐mediated proliferation of EC‐GI‐10 + ERα cells, EC‐GI‐10 + ERβ cells. The proliferative activity was evaluated as a ratio (%) compared to that of controls (no treatment with either estradiol or ICI 182780 for 72 h). Mean ± SD (n = 6). *P < 0.05 versus controls (Kruskal–Wallis test and Scheffe test).

Effects of ERα or ERβ expression on estradiol, propyl‐pyrazole‐triol and diarylpropionitrile‐mediated cell proliferation in EC‐GI‐10 cells

The number of EC‐GI‐10 + ERα cells was significantly decreased following the treatment with PPT (1 μmol/L), but not with estradiol (1 μmol/L) or DPN (1 μmol/L) compared to that administered with vehicle control cells (Fig. 3C, left). Both estradiol (1 μmol/L) and ERβ specific agonist DPN (1 μmol/L) treatments significantly increased the cell proliferation of EC‐GI‐10 + ERβ cells when compared to that administered with vehicle control cells (Fig. 3C, center). The estradiol‐mediated cell proliferation of EC‐GI‐10 + ERβ cells was significantly suppressed (P < 0.001) by the addition of ICI 182780 (100 pmol/L) and reached the control level. There was no significantly difference in the number of EC‐GI‐10 + CMV cells treated with estradiol, DPN or TTP (Fig. 3C, right). Estradiol‐mediated cell proliferation was detected in EC‐GI‐10 + ERβ cells and was significantly induced following the treatment with 1 nmol/L to 1 μmol/L estradiol (the difference of groups; Kruskal–Wallis, P = 0.0004, NC v 1 nm to 1 μm have difference in Scheffe test; Fig. 3D, right). Estradiol decreased the cell proliferation of EC‐GI‐10 + ERα in the concentration of 1 pmol/L to 1 μmol/L, but this increment did not reach statistical significance (Fig. 3D, left)

Discussion

Freedman et al. (27) report that menopausal hormone therapy is significantly associated with lower risk of head, neck and ESCC in the National Institutes of Health‐American Association of Retired Persons (NIH‐AARP) Diet and Health Study.27 Bodelon et al. also report the significant association between hormone replacement therapy and the risks of developing ESCC in postmenopausal women enrolled in a Women's Health Initiative (WHI) and observational study.28 They both demonstrate that the women who took both estrogen and progestin had lower risk of developing ESCC than those who took a placebo, but this association was not detected when women took estrogen alone.28 Further investigations regarding hormone receptor status, including progesterone receptor in ESCC, are certainly required for clarification, but it has become important to examine the details of estrogenic actions, especially the status of ER in development and behavior of ESCC patients.

Enmark et al. report the presence of two different isoforms of ER, ERα and β, in many types of human tissues.5 Subsequent studies confirm that ERα and ERβ are not only expressed in classical estrogen target tissues but are also rather widely distributed in humans. Taylor et al. report the presence of ERα and ERβ in normal esophagus squamous epithelium.29 Nozoe et al. report the presence of ERα and ERβ in ESCC patients.13 Kalayarasan et al. report that the status of ERβ is correlated with aggressive behavior in ESCC patients,15 but different results have been also reported.17 Therefore, the clinical and biological significance of ERα and ERβ in ESCC has not been clarified.

In the present study, we examined the nuclear immunoreactivity of ERα and ERβ in both squamous cell carcinoma and non‐neoplastic squamous cell epithelium of the esophagus. A relatively high level of ERα immunoreactivity was detected in the nuclei of non‐neoplastic basal layer cells in normal esophageal mucous. In mammary glands, Khan et al. report increased ERα immunoreactivity in normal epithelium obtained from tumor‐bearing breasts compared to non‐tumor bearing breast.30 Lawson et al. also report that ERα expression is higher in the breast tissue of women from a population at high risk of breast cancer compared with that in the tissue of women associated with a relatively low risk of the disease development.31 It is also interesting to note that Zhai et al. report the loss of ERα expression in the advanced stages of cervical squamous cell carcinoma progression.32 In the present study, a relatively high level of ERα immunoreactivity was also detected in non‐neoplastic esophageal mucosa bearing ESCC compared to that in concomitant or adjacent carcinoma cells. Therefore, in human squamous mucosa, estrogen may help to maintain normal cell cycle or exert a protective effect upon epithelial cells through the ERα. Further investigation is necessary to clarifying this interesting hypothesis, for example through comparison of ERα status of morphologically normal esophageal mucosa in tumor bearing and non‐tumor bearing subjects. However, it is also true that the status of ERα immunoreactivity in ESCC was by no means associated with any of the clinicopathological variables of the patients examined in this study, including their clinical outcome. In contrast, ERβ status of carcinoma cells was significantly associated with unfavorable clinical outcome of the patients and ERβ status in carcinoma cells also turned out to be an independent unfavorable prognostic factor for the patients (determined using multivariate analysis). Therefore, it has become important to examine the effects of estrogen signals mediated through ERβ in these patients to clarify the possible involvement of estrogens in the biological behavior of ESCC.

Ivanova et al. report on the difference between men and women in the modes of actions of ERβ in lung cancer.33 Therefore, we initially postulated that variations in expression of ER were due to the differences in the prevalence of gender in ESCC. However, in the present study there were no differences in the status of ER according to the gender of the patients. Therefore, the gender differences in the prevalence of ESCC might be due to differences in the lifestyles of male and female patients (e.g. drinking and smoking), but further investigations are required for clarification.

In the present study, ERβ was detected in the nuclei of ESCC, as in Nozoe et al. The differences between the results of the present study and those of Nozoe et al. could be due to the ERβ antibodies used and the evaluation method of ERβ immunoreactivity. Nozoe et al. determined that ERβ nuclear staining in at least 50% of tumor cells were scored as positive for overexpression, based upon a study of lung cancer by Wu et al.13, 34 However, we defined ERβ nuclear positive immunoreactivity or overexpression as the cases with an H score of more than 250. The validity of this definition is discussed in the Materials and Methods section. In breast carcinoma, the difference of the cut‐off points of the ER status is well‐known to be related to the prognosis of individual patients.35, 36 The results of survival analyses of breast carcinoma patients are also known to be different depending on the evaluation method of the ER (e.g. proportion score and intensity score).35 Further examinations using different ERβ antibodies and evaluation methods are obviously required to clarify the ERβ and prognosis in ESCC.

According to Matsuoka et al. and Ueo et al., estrogen prevents cell proliferation of primary ESCC cells, which are reported to be associated with the presence of ER.3, 4 ERβ was not identified in the 1980s and, therefore, results from these previous reports3, 4 could not clarify whether estrogenic signals were mediated in ESCC through ERα, ERβ or both. When both ERα and ERβ are present in the cells and bound to estrogen, estrogen signals through ERβ are generally considered to inhibit ERα‐dependent transcription.37 For instance, estrogen‐dependent cell proliferation is reported to be inhibited by transfection of ERβ in breast carcinoma cells.38 However, several studies state that estrogenic action through ERβ signaling induces rather than suppresses the cell proliferation of carcinoma cells.8, 39, 40 For instance, Teng et al. demonstrate that the treatment of ERβ specific agonist DPN significantly increased cell proliferation in primary urothelial cells, which predominantly expressed ERβ.10 They also report that estrogenic signals mediated through ERβ predominantly induce G1/S transition in primary urothelial cells.41 Hershberger et al. report ERβ‐dependent cell proliferation through both genomic and non‐genomic pathways in lung carcinoma cell line.42 In the present study, the treatment of PPT, the specific ERα agonist, significantly decreased the number of EC‐GI‐10 + ERα cells and that of estradiol, and DPN, the specific ERβ agonist, significantly increased the number of EC‐GI‐10 + ERβ cells. Results of the present study clearly indicate that estrogenic actions through ERβ were predominant in ESCC cells compared to those through the ERα. Therefore, estrogen might also play a pivotal role in the cells in which estrogenic signals are mediated predominantly through ERβ, such as ESCC cells. Interestingly, TGFα, amphiregulin or HB‐EGF, all of which are considered EGFR activators, are induced by estradiol treatment in ERα negative/ERβ positive breast carcinoma cell line43 or ERβ dominantly expressed HNSCC cells.44 EGFR is also reported to be detected in many ESCC cells,45, 46 and is even considered as a target molecule for a potential target specific therapy in ESCC patients.47 Therefore, estrogenic effects, including an induction of its target genes, might facilitate the proliferation of ESCC cells through the activation of EGFR signals in ESCC, but further investigations are required for clarification.

In summary, we demonstrated ERα and ERβ expression using immunohistochemistry in human ESCC. The status of nuclear ERβ immunoreactivity in carcinoma cells turned out to be an unfavorable independent prognostic factor in ESCC patients. Results of our immunohistochemical and in vitro studies clearly demonstrate that ESCC is an estrogen‐dependent human malignancy, as in other human cancers. ERβ might serve as a novel target molecule for ESCC patient therapy.

Disclosure Statement

The authors have no conflict of interest.