Non-small cell lung cancer (NSCLC) accounts for 75–80% of lung cancer patients and its dismal survival rate has been improved only marginally in the past 2 decades. The 5-years survival rate of patients with NSCLC is only about 15%.1 This illustrates the need for more effective strategies for early diagnosis and chemoprevention. Identification of useful biomarkers for early detection and predication of patient survival are important to determine a better clinical treatment and our findings may be helpful for the clinical outcome of NSCLC patients.
Previous studies have indicated that ER β was detected in all tested lung cancer cell lines and lung tumor tissues compared to ER α that was partially detected in lung cell lines and lung tumor tissues.2, 3 We considered that ER α may play a more important role than ER β in lung cancer. Our study was focused on ER α (ER). The ER is located on chromosome 6q25.14, 5 and belongs to the super family of transcription activators.6, 7 Its protein product is a transcription factor that regulates the expression of estrogen responsive gene by binding to a specific DNA sequence found in their regulatory regions. As a mediator of estrogen hormone action, ER is involved in many physiological processes.8 Loss or downregulation of ER expression in breast,9, 10 ovary11, 12 and prostate cancers13 has been documented frequently. Low ER expression was also associated with a poor prognosis for effective endocrine therapy.14, 15, 16 Mechanisms regulating the expression of ER are poorly defined, unfortunately, and no mutation or other gross structure alteration of the ER gene in lung cancer has been reported that could clarify the mechanism. Our previous reports indicated that DNA adduct levels may act as a risk marker of lung cancer and the adduct levels in female nonsmoking lung cancer patients were significantly higher than those of male nonsmoking patients.17, 18 We suggested that ER might play a role in the gender difference of DNA adduct levels, and it may be helpful to understand why female lung cancer had a higher susceptibility to DNA damage derived from environmental carcinogens. We hypothesized that gender difference in loss of ER expression may affect lung cancer risk. Additionally, ER inactivation by promoter hypermethylation may alter antiestrogen therapeutic response that would result in different prognostic values between male and female lung cancer patients.
Loss of ER expression has been associated with aberrant 5′CpG island hypermethylation in breast cancer cell lines and their tumor tissues.19, 20 The cause of aberrant hypermethylation in lung cancer and the clinical determinants associated with such hypermethylation have not been well defined. Observations in animal model of lung cancer have indicated that ER gene hypermethylation was contributed partially by carcinogenic insults that induced tumor formation. In this model, ER gene in spontaneous and radiation-induced tumors was much more likely to be methylated than that in tumors induced by the tobacco-derived carcinogen, NNK.21, 22 This data suggests the possibility that ER hypermethylation may be also associated with specific clinical characteristics in lung tumors. Stabile et al.2 indicated that estrogen signaling played a biological role in lung tissues and that estrogen could potentially promote lung cancer, either through direct actions on preneoplastic or neoplastic cells or through indirect action on lung fibroblasts. It was conceivable that loss of ER expression by promoter hypermethylation may modulate lung tumor progression and even have different prognostic values between male and female NSCLC patients.
Material and methods
Lung tumor cancer tissues obtained by surgical resection and adjacent normal lung tissues were collected respectively from 123 NSCLC lung cancer patients (58 male smokers, 33 male nonsmokers, 32 female nonsmokers) admitted to Taichung Veterans General Hospital, Taichung, Taiwan between 1997–2001. Patients were followed up from January 1997 with a combination of active follow-up and record linkage to the death certificates. Their censoring data was assigned to be December 31 2001, 6 months before our search of the vital statistics registry, to follow for a possible delay of entry of the death certificates into the registry. No lung tumors for female smokers were available for our study because <10% of female lung cancer patients were smokers in Taiwan, and most cases could not be carried out with surgical therapy because they were diagnosed to be advanced lung cancer. The histology, types and stages of tumors were determined according to the WHO classification method (WHO, 1982). Information on smoking history of lung cancer patients was obtained from hospital records and used to categorize patients into smoking and non-smoking groups. Tissues were stored at −80°C after resections until used.
Genomic DNA was isolated from lung tumor tissues and A549 cells by conventional phenol-chloroform extraction and ethanol precipitation. Hypermethylation status of the ER promoter region was determined by a bisulfite genomic sequencing protocol and methylation-specific PCR (MSP).23 The bisulfite-modified DNA was amplified by using primers specific for the methylated and unmethylated ER sequence, respectively.23, 24 The ER CpG islands were amplified from bisulfite-treated genomic DNAs by PCR using the same pairs of primers described earlier.23 Amplified products were sequenced using the ABI sequencing system (Applied Biosystem, Foster City, CA).
The expression of ER in lung cancer tumors and A549 cells was examined by RT-PCR. Total RNA from lung cancer tissues was extracted by Trizol regent (Life Technologies, Grand Island, NY) according to the manufacturer's recommendations. First-strand cDNA synthesis with oligo-dT primers was carried out using Superscript II (reverse transcriptase). The size of amplified ER fragment was identified to be 782 bp with the primers described as below: sense, GCAATGACTATGCTTCAGGCTACC; antisense, AGGCACACAAACTCCTCTCCC. GAPDH and 18s rRNA were used as an internal control for RT-PCR.
A549, a human lung cancer cell line, was grown in RPMI-1640 supplemented with 10% FBS, penicillin (100 U/ml), streptomycin (100 mg/ml) and 5% glucose. Under estrogen-depleted conditions, cells were grown in estrogen-depleted medium, consisting of phenol red-free RPMI-1640 (Sigma, St. Louis, MO), 10% charcoal-stripped FBS, penicillin (100 U/ml), streptomycin (100 mg/ml) and 5% glucose as described previously. Cells were grown in a 37°C humidified incubator with 5% CO2.
In vitro treatment of lung cells with 5-AZA-dCyd
Cells were plated with a density of 3 × 105/100 cm2 dish and incubated for 24 hr. Culture media were replaced with a media containing 10 nM 5-AZA-dCyd (5-aza-2-deoxycytidine; Sigma, St. Louis, MO), dissolved in 50% acetic acid and storage at −80°C). After an incubation of 5 days, treated cells were subjected to molecular analyses.
Effects of 17-β estradiol on ER promoter hypermethylation of lung cancer cells
The A549 cells were pre-cultured for about 2 days in estrogen-depleted medium followed by a 2-hr treatment in the presence or absence of 5 nM 17-β estradiol. Histones of A549 cells were extracted according to the procedure of Yoshida et al.25 Cells (2 × 106) were collected and washed with ice-cold PBS, suspend with 1 ml ice-cold lysis buffer (10 mM Tris-HCl, 50 mM NaHSO3, 1% Triton X-100, 10 mM MgCl2, 8.6% sucrose, pH 6.5) and then lysed by being passed through a syringe with 22-gauge needle. The nuclei were collected by a centrifugation at 12,000g for 5 min, washed 3 times with cold lysis buffer, and 1 time with 10 mM Tris-HCl, 13 mM EDTA, pH 7.4, successively. The pellet was suspended in 100 μl ice-cold H2O by vortexing and concentrated H2SO4 was added to the suspension to give a concentration of 0.4 N. After incubation at 4°C for at least 1 hr, the suspension was centrifuged for 5 min at 15,000g and the supernatant was mixed with 1 ml of acetone. After an overnight incubation at −20°C, the coagulated material was collected by micro-centrifugation and air-dried. This acid-soluble histone fraction was dissolved in 30 μl of H2O and quantitated using a protein assay kit (Bio-Rad Laboratories, Hercules, CA). Acid-extracted histones from A549 cells treated with or without 17-β estradiol were resolved by 15% SDS-PAGE, transferred to PVDF (Millipore, Bedford, MA), blocked with 3% non-fat milk, and probed with acetylated H3 and H4 type-specific antibodies as indicated (Upstate Biotechnology Inc., Charlottesville, VA) for Western blotting. Proteins were visualized using a goat anti-rabbit secondary antibody conjugated to horseradish peroxidase and an enhanced chemiluminescence detection system. Gels were stained with Coomassie Brilliant Blue R-250 (Bio-Rad), dried and photographed.
Fisher's exact or χ2 test was applied for statistical analysis. For survival data, statistical differences were analyzed using the log-rank test. Survival curves were plotted using the Kaplan-Meier method and variables related to survival were analyzed using Cox's proportional hazards regression model with SPSS software (SPSS Inc., Chicago, IL). A p < 0.05 was considered to be statistically significant.
Comparison of ER hypermethylation between paired lung tumor and adjacent normal lung tissues
Our study recruited 123 lung cancer patients (58 male smokers, 33 male nonsmokers, 32 female nonsmokers). ER mRNA expression and ER hypermethylation were detected by RT-PCR and MSP, respectively. Representative results of RT-PCR of lung tumors were shown in Figure 1a and MSP results of lung tumor and adjacent normal lung tissues from lung cancer patients were shown in Figure 1b. The MSP data were confirmed by bisulfite-treated DNA sequencing (Fig. 1c). Our data indicated that ER hypermethylation was detected in lung tumor tissues (54%, 66/123) but not in any adjacent normal lung tissues (0%, 0/75) (Fig. 1b, Table I). We suggest, therefore, that ER hypermethylation may be associated with lung tumorigenesis.
Table I. Relationships between Clinical Parameters and Estrogen Receptor Hypermethylation in 123 NSCLC Patients
Relationships between ER hypermethylation and clinical pathological parameters of lung cancer patients
Among studied clinico-pathological parameters including age, gender, tumor type, differentiation grade, tumor stage, cigarette smoking status, T and N values, only gender factor was significantly correlated with ER hypermethylation (Table I). Male lung cancer patients had a higher frequency of ER hypermethylation than female lung cancer patients (p = 0.01). Male cancer patients that smoked had the highest frequency of ER hypermethylation, followed by male and female nonsmokers. It is notable that a significant difference was observed between male and female nonsmokers (p = 0.03) but not between male smokers and nonsmokers (Table II). Our results suggest that the impact of gender difference on ER hypermethylation was more profound than cigarette smoking.
Table II. Prevalence of ER Hypermethylation in Lung Tumors by Gender and Smoking Status
Chi-square test for categorical variables (nonsmoking males vs. females), p = 0.03.
Chi-square test for categorical variables (nonsmoking vs. smoking males), p = 0.98.
Role of E2 in ER promoter methylation in lung cancer A549 cells
To verify whether E2 was involved in the elimination of ER hypermethylation and ER mRNA expression, a lung cancer A549 cell with ER hypermethylation and absence of ER mRNA expression was used. RT-PCR and MSP data indicated clearly that ER mRNA was only detectable in A549 lung cancer cells after a treatment with 1 nM and 5 nM of E2 (Fig. 2a). Although the methylation band disappeared after E2 treatment, the unmethylated band remained unaltered (Fig. 2b). Additionally, Western blot data showed that acetylated histone 3 (H3) and histone 4 (H4) of chromatin were increased significantly after a 40-min treatment of 5 nM E2 (Fig. 2c). Our data indicates that E2 may promote H3 and H4 acetylation to attenuate ER promoter methylation and cause ER re-expression. We suggest that a lower frequency of ER hypermethylation in female lung cancer patients compared to male patients may be due to a relatively higher E2 concentration in the blood of females.
Correlation between ER hypermethylation and ERmRNA expression
To verify whether promoter hypermethylation was linked with ER transcription silencing, lung cancer A549 cells with ER hypermethylation were treated with demethylating agent 5-AZA-dCyd to understand whether the hypermethylation of CpG islands in ER exon 1 was responsible for the gene transcription silencing. Our RT-PCR data showed that ER mRNA was restored in A549 cells after a treatment with 5-AZA-dCyd. MSP data showed the absence of ER promoter hypermethylation in 5-AZA-dCyd treated-A549 cells (Fig. 3), but ER hypermethylation was observed in the parental and vehicle-treated cells. Our results suggest strongly that the location of ER promoter hypermethylation detected by the primer used in our study plays a crucial role in ER transcriptional regulation. The correlation of ER hypermethylation with ER mRNA was statistically analyzed in our lung tumor study. The representative results of RT-PCR and MSP for ER mRNA expression and hypermethylation are shown in Figure 1a. Direct sequencing of ER CpG islands was carried out to confirm the above results. For all patients studied, ER mRNA expression was correlated significantly with ER hypermethylation (Table I, p = 0.01). Our results suggest that ER hypermethylation may be responsible partially for ER transcriptional silencing in lung cancer patients.
Prognostic value of ER hypermethylation in NSCLC patients
The association of ER hypermethylation with various clinico-pathological parameters with patients' survival was statistically investigated by univariate analysis (Table III). Results showed that several parameters, including ER hypermethylation, gender, tumor stage and N value, were associated independently and significantly with patient survival (p = 0.0058 for ER hypermethylation, p = 0.0125 for gender, p = 0.0081 for tumor stage, p = 0.0016 for N value). Patients with parameters including ER hypermethylation, Stage III, N1 and N2 nodal micrometastasis and males had a shorter survival than those with Stage I and II, N0 and females. Kaplan-Meier analysis showed that patients with ER hypermethylation had a poorer prognosis than those without ER hypermethylation (Table III). Additionally, ER hypermethylation as a valuable prognostic marker was only feasible for male lung cancer patients (Fig. 4b, p = 0.0044), but not for female lung cancer patients (Fig. 4c, p = 0.6369). Our results supported the hypothesis that there was a gender difference in the prognostic values of loss of ER expression by hypermethylation. Moreover, Cox regression analysis data indicated that patients with ER hypermethylation had a significantly shorter survival than those with the absence of ER hypermethylation (p = 0.007, Table IV). Among all study cases, the risk ratio (RR) of patients with ER hypermethylation was nearly 2.0-fold of patients in the absence of ER hypermethylation, which was quite similar to that of tumor stage (RR = 2.1) to make ER hypermethylation as well as tumor stage be an independent prognostic factor.
Table III. Univariate analysis of Influences of Clinical Characteristics on Overall Survival Duration of NSCLC Patients
Log-rank p-values for categorical variables were statistically analyzed by Kaplan-Meier test.
Squamous cell carcinoma
Table IV. Cox Regression Analysis of Various Potential Prognostic Factors in NSCLC Patients1
Adjusted for age, stage, gender and smoking status.
ER hypermethylation was responsible for ER gene inactivation not only in breast cancer,19, 20, 26 but also in adult acute myeloid leukemia,27 endometrial cancer,28 prostate cancer29 and hepatocellular carcinoma.30 The prognostic value of ER hypermethylation in lung cancer, however, remains unclear. To our knowledge, this is the first study to show that ER hypermethylation may act as an unfavorable prognostic factor of NSCLC patients. In our study, we also observed that ER hypermethylation was occurred in lung tumor, but not in adjacent normal part (Fig. 1). Additionally, our cell line experiment indicated that the exon 1 of ER promoter hypermethylation was responsible for the ER transcriptional silencing (Fig. 3). These results seem to show that ER promoter hypermethylation in exon 1 CpG islands may play a role in lung tumorigenesis.
The prevalence of ER hypermethylation observed in lung cancer patients that smoke seemed to give rise to the speculation that tobacco smoke could affect ER hypermethylation in NSCLC.21 Previous studies have indicated that the prevalence of some specific gene hypermethylation (i.e., p16 and RASSF1A), was increased by the duration of cigarette smoking or an early age at starting smoking.31, 32 A previous study indicated, however, that NNK-induced rodent and rat lung tumors had relatively lower frequencies of ER hypermethylation compared to those induced by X-ray, plutonium and spontaneous tumors.22 For a few primary human lung tumors used in the same study (n = 46), there was no difference in the prevalence of ER hypermethylation with respect to gender, tumor type or tumor stage. Moreover, the frequency of ER hypermethylation in lung cancer patients that smoke was significantly higher than those of nonsmoking patients. There was no difference in ER hypermethylation between male smokers and nonsmokers (Table II) in the 123 lung tumors tested in our study. Our results suggest that other environmental factors may play a more important role than cigarette smoking in the occurrence of ER hypermethylation. The precise mechanism should be further defined.
We found a significant gender difference in ER hypermethylation. The frequency of ER hypermethylation in male lung cancer patients was significantly higher than that in female lung cancer patients (Tables I, II). A previous study for ER mRNA expression on fewer lung cancer patients, including 13 females and 13 males, has also found that ER mRNA expression was more prevalent in females than in males.33 This is consistent with our results. Furthermore, we hypothesized that E2 may play a role in ER hypermethylation based on the study of Sakai et al.34 showing that a treatment of E2 and progesterone increased significantly the levels of acetylated histone 3 and 4 of chromatin in human endometrial stromal cells. A consistent result was observed in our study indicating that ER demethylation and ER mRNA re-expression occurred simultaneously in A549 lung cancer cell after a treatment with E2 (Fig. 2a,b). Furthermore, acetylated histone 3 and 4 of chromatin were gradually increased by E2 treatments (Fig. 2c). Our present data showed the occurrence of ER demethylation after estrogen treatment in A549 lung cancer cell, but the detail mechanism needs further investigation. It was conceivable that female lung cancer patients had relatively higher levels of E2 than male patients to protect ER gene inactivation through ER hypermethylation. In our study, ER hypermethylation occurred in about 60% of male lung cancer patients, but in only 34% of female patients. Comparing with that in male lung cancer patients, a higher prevalence of ER hypermethylation was also observed in hepatocellular carcinoma (62%, 53/85)30 and prostate cancer (95%, 36/38).35 Both cancers were well known to occur in men more frequently. Additional experiments will be carried out to elucidate the role of testosterone in ER hypermethylation of lung cancer patients.
An early transfection report indicated that a significant cell growth inhibition and cell apoptosis were observed in ER transfected breast cancer cell compared to the parental cells.36 Approximately 60% of proliferating cells were seen in the mammary gland without ER. Some investigators have observed increased doubling times after they introduced ER into ER negative cells.37 This role played by ER α is very similar the role of P14ARF tumor suppressor, which binds to p53 and mdm2 and protects p53 from being downregulated by mdm2.38 The involvement of ER transcription silencing by promoter hypermethylation in lung tumor progression was at least in part mediated through an alteration of p53-mdm2 feedback regulation.
ER gene inactivation has been correlated closely with the resistance to anti-hormonal treatments. Accumulated data indicated that a high level of ER α could increases endocrine-base therapy response in the control of ovarian, endometrial and breast cancers.39, 40, 41 A recent study reported by Stabile et al.2 indicated that E2 produced a proliferative response in vitro in normal lung fibroblasts and cultured non-small cell lung tumor cells. They also demonstrated that E2 stimulated transcription of an estrogen response element-luciferase construct transfected in lung tumor cell lines. These results suggested that estrogen signaling plays a biological role in normal lung fibroblasts and lung cancer cell lines. More importantly, another study showed that the proliferation of lung cancer cells stimulated by E2 was reduced significantly by the treatment of tamoxifen.3 Some clinical studies have pointed out that female gender exerts a significant positive effect on survival after lung surgical resection therapy for early stages of NSCLC patients and favorable response rates and survival times were obtained in the treatment of antiestrogen drug, tamoxifen, in the combination with cisplatin and etoposide to advanced NSCLC patients.42, 43 We observed different prognostic values of ER hypermethylation for male and female lung cancer patients. A positive ER expression in female lung cancer patients may indicate a favorable response of surgical resection and antiestrogen therapy. Antiestrogens may have an enormous value in treating or preventing lung cancer, especially for female lung cancer patients who were frequently ER-positive.