Promoter methylation of sFRP5 in patients with ovarian clear cell adenocarcinoma


Prof. Shwu-Fen Chang, PhD, Graduate Institute of Medical Sciences, Taipei Medical University, 250 Wu-Hsing Street, Taipei 110, Taiwan. Tel.: +886-02-27361661; fax: +886-02-23778620; e-mail:


Eur J Clin Invest 2010; 40 (4): 310–318


Background  Specific tumour suppressor genes with promoter methylation in ovarian clear cell adenocarcinoma (OCCA) can be one important epigenetic mark distinguishing OCCA from ovarian serous adenocarcinoma (OSA), benign endometriotic cysts and normal ovarian epitheliums.

Materials and methods  Five OCCA cell lines, 63 cancer tissues (48 OCCA and 15 OSA), 10 benign endometriotic cysts and five normal ovarian epitheliums were analysed by methylation-specific PCR using pooled DNAs to determine the methylation status of the promoter of the target genes, including genes for secreted frizzled-related proteins (sFRP1 to 5), adenomatous polyposis coli (APC), retinoblastoma protein 1 (Rb1), breast cancer 1 gene (BRCA1), p14ARF, p15INK4b, p16INK4a and survivin. Methylation frequencies of identified targets were further analysed with individual DNA samples.

Results  The sFRP5 promoter was significantly methylated in all OCCA cell lines, with 64·6% in OCCA tissues compared with 13·3% in OSA, and 0% in benign endometriotic cysts and normal ovarian epitheliums (P < 0·0001). With a median follow-up of 44 months, the expected 5-year overall survival (OS) for patients with methylated sFRP5 promoter were significantly worse than for those with unmethylated sFRP5 (52% vs. 88%, P = 0·03). After adjusting for age, stage, and residual disease after primary surgery, patients with unmethylated sFRP5 promoter had an independent good prognostic factor in OS (P = 0·017).

Conclusion  The high percentage of promoter methylation in the sFRP5 gene in OCCA indicates its importance in the development of OCCA and is a potential useful marker for prognoses and target for treatment of OCCA.


Ovarian clear cell adenocarcinoma (OCCA) has been recognized as a distinct histological type of cancer and its frequency among epithelial ovarian cancers is estimated to be less than 5–10% [1,2]. However, the overall frequency of OCCA is higher in Taiwan [3]. OCCA clinically differs from other epithelial ovarian cancers because of its de novo resistance to platinum-based chemotherapy, and shows a poor prognosis [3–5].

The molecular pathogenesis of OCCA is still unclear, although several efforts have been made. Hepatocyte nuclear factor-1β was up-regulated in OCCA cells compared with that in non-OCCA cells, and was essential for the survival of patients [6]. Lower TP53 and cyclin A levels but higher p21 and cyclin E levels were detected in OCCA compared with other epithelium ovarian cancers, and were thought to involve in the carcinogenesis of OCCA [7]. Silencing of Wilms tumour suppressor 1 sense and antisense genes by promoter methylation distinguished OCCA from ovarian serous adenocarcinoma (OSA) [8]. In addition, loss of heterozygosity in p53 gene was detected in non-OCCA epithelium ovarian cancer [9].

Epigenetic events become important in the development of many cancers. Hypermethylation of the CpG islands within the regulatory region of tumour suppressor genes (TSGs) is one of the earliest and most frequent alterations, which results in gene silencing may confer a growth advantage for tumour cells [10]. Cellular events affected include DNA repair, cell cycling, adherence, apoptosis and detoxification [11]. The distinct patterns of DNA methylation among different tumours may be a useful signature for diagnosis and prognosis [12].

Aberrant activation of Wnt signalling is associated with many human cancers [13]. Secreted frizzled-related proteins (sFRPs) are extracellular and antagonize the Wnt signalling either by interacting with Wnt proteins to prevent them from binding to the frizzled receptors (Fz) or by forming nonfunctional complexes with Fz. Accumulating evidence suggests that sFRPs act as tumour suppressors because their expression is frequently silenced in cancer by promoter hypermethylation and that they may serve as epigenetic tumour biomarkers.

There are five known members of the sFRP family – sFRP1 to sFRP5 – each with promoter hypermethylation associated with various cancers. Takada et al. reported that two of 17 primary ovarian cancers had methylated sFRP1 promoter, meanwhile, one of three OCCAs had trace amounts of methylated sFRP1 promoter [14]. Loss of sFRP1 has been reported recently as an early aberrant molecular event in clear cell renal cell carcinoma [15]. Recent analyses have identified comparable genetic alterations in tumours with similar differentiation, regardless of the organ origin [16]. Furthermore, promoter hypermethylation at sFRP1, sFRP2, SOX1 and LMX1A genes was reported to correlate with the recurrence and overall survival of ovarian cancer [17]. However, promoter methylation of these genes has not been analysed in OCCA. We therefore hypothesized that promoter methylation of the Wnt antagonist genes may possibly contribute to the pathogenesis and progression of OCCA. We applied methylation-specific polymerase chain reaction (MS-PCR) analysis to reveal the promoter methylation profile of sFRP1, 2, 4, 5 genes and seven other associated genes (APC, BRCA1, P14ARF, P15INK4b, P16INK4a, Rb1 and survivin) in OCCA.

Materials and methods

Cell lines and cultures

OCCA cell lines including HAC-2, KK, RMG-I, RMG-II (obtained from Professor Shiro Nozawa of the Department of Obstetrics and Gynecology, Keio University; Professor Yoshihiro Kikuchi of the Department of Obstetrics and Gynecology, National Defense Medical College; and Dr Masato Nishida from Kasumigaura National Hospital), and ES-2 cells (obtained from the American Type Culture Collection, ATCC) were used in this study. Two immortalized cell lines, OSE2a (non-tumorigenic) and OSE2b-2 (tumorigenic) (kindly given by Professor Hironori Tashiro of the Department of Gynecology, Kumarmato University), were also used. All cells were maintained as described previously [18].

Patients and tumour sample collections and genomic DNA extraction

This project has been approved by the Institutional Review Board of Cathay General Hospital (CGH). Tissue samples were obtained from surgical specimens with the informed consent of patients at the CGH Fresh frozen tumour tissues of patients with OCCA, other types of ovarian cancers, benign ovarian cysts and normal ovarian epithelium were collected following the regulations approved by IRB of CGH, and stored at −80 °C. Sixty-three primary human epithelial ovarian carcinoma samples, comprising 48 OCCA and 15 OSA tissues, 10 benign endometriotic cysts, and five normal ovarian epithelial tissues were collected from the gynaecological cancer centre in CGH between 1994 and 2002. The median age of five patients with normal ovarian epithelial tissues was 52 years old (range: 50–58). Among them, 26 samples including 15 OSA and 11 OCCA were frozen tissues taken only from cancerous regions at the time of surgery, and were confirmed pathologically to have high neoplastic cellularity (> 66%) in each sample. The other 37 cancer tissues were micro-dissected from paraffin-embedded sections. The pathological reviews were undertaken by two gynaecological pathologists (S.-H. Huang, M.-C. Lin). Genomic DNA was extracted from these specimens or cells using the QIAamp tissue kit (Qiagen, Valencia, CA, USA) amplicons following the instructions of the manufacturer.

Pooled DNA approach

For rapid identification of potential targets with aberrant methylation, promoter methylation status of 11 genes (sFRP1, sFRP2, sFRP4, sFRP5, APC, BRCA1, P14ARF, P15INK4b, P16INK4a, Rb1, survivin) was determined on pooled DNA samples with MS-PCR. Pooled DNA mixtures from 11 OCCA fresh frozen tissues, 37 OCCA paraffin block tissues, 15 OSA frozen tissues, 10 benign endometriotic cysts, and five normal ovarian epithelial tissues were made to form four sample groups. Having identified potential markers, frequencies of methylation were further analysed in individual samples.

Sodium bisulfite genomic treatment and MS-PCR

The DNA was sodium bisulfite modified using EZ DNA methylation kit (Zymo Research, Orange, CA, USA), purified and PCR amplified with Gold Tag DNA polymerase (PE Applied Biosystems, Foster City, CA, USA) using gene-specific primers listed in Table 1 as described previously [18]. Bisulfite-modified, SssI (New England Biolabs, Beverly, MA, USA)-treated normal lymphocyte DNA served as the methylated control; bisulfite-treated normal lymphocyte DNA was the unmethylated control. PCR product was analysed on 3% agarose gel.

Table 1.   Primer list
Gene nameSense (5′→3′)Antisense (5′→3′)
Primer for MS-PCR
Primer for real-time-PCR

5-Aza-2-deoxycytidine treatment and reverse transcription-real time PCR analysis

Cells were treated with 10 μM of 5-Aza-2-deoxycytidine (5-aza-2-dC; Sigma, St. Louis, MO, USA), and renewed every 24 h. Reverse transcription-real time PCR analysis (RT-qPCR) was used to measure the sFRP5 mRNA in response to the treatment (15 days). A constitutive expressed GAPDH was used as an internal control. The qPCR reaction was performed in an ABI Prism 7300 Sequence Detection System (Applied Biosystems) with SYBR Green master mixture and primers listed in Table 1. The condition was as follows: 2 min at 50 °C, 10 min at 95 °C, and a two-step cycle at 95 °C for 15 s, 60 °C for 1 min for 40 cycles with an additional dissociation curve. The interpolated number (Ct) of cycles to reach a fixed threshold above background noise was used to quantify amplification.

Clinicopathological characteristics of the patients

The clinicopathological characteristics of the patients were collected by reviewing medical records, age, FIGO stage, optimal cytoreduction, adjuvant therapy and histological type of cancer. The survival time was defined as the duration from the time of diagnosis to death or last contact.

Statistical analysis

The promoter methylation of sFRP5, APC, BRCA1 and survivin genes among OCCA, OSA, benign endometriotic cysts and normal ovarian epithelial tissues was compared. Survival time was considered the primary end point and was defined as that from the date of diagnosis to the date of death or last contact. Survival information was available for all patients. Survival curves were generated using the Kaplan–Meier method. Differences in survival curves were calculated using the log rank test. A P-value of < 0·05 was used to determine statistical significance. Cox’s univariate and multivariate regression analysis was used to evaluate prognostic factors for survival.


The promoter methylation profile of tested genes in pooled DNA from OCCA

To facilitate the investigation, we first analysed the promoter methylation of tested genes in the pooled DNA from clinical samples. Promoters of APC, and sFRP1, 2, and 5 in pooled samples were found methylated in all tested ovarian cancers. BRCA1 promoter was methylated in OSA and benign ovarian cysts. Survivin promoter in pooled samples was methylated in all tested ovarian cancers, benign endometrioid cysts and normal ovarian epithelial tissues. However, DNA methylation was not detected in the promoter regions of, RB1, p16INK4a, p15INK4b and p14ARF genes in all tissue samples.

The promoter methylation profile of the sFRP gene family in OCCA

To investigate the promoter methylation of the sFRP gene family in OCCA compared with that in OSA, benign endometriotic cysts, and normal ovarian epithelial tissues, we analysed the methylation profiles of sFRP1, 2, 4 and 5 gene promoters in ES2, KK, RMG-I, RMG-II, and HAC2 five OCCA cell lines, two immortalized OSE2a (non-tumorigenic) and OSE2b-2 (tumorigenic) cell lines, tissues of OCCA, OSA, benign endometriotic cysts and normal ovarian epithelial tissues. Results showed that sFRP1 promoter methylation was 33·3% (12/36) in OCCA samples. We were unable to determine the promoter methylation profile of 12 DNA samples from paraffin-block tissues, which may be attributable to severe DNA degradation to impede the MS-PCR with sFRP1 primers (Fig. 1). Among the five OCCA cell lines tested, only RMG-I and RMG-II showed hypermethylation in the promoter of sFRP1. The prevalence of promoter methylation of sFRP2 and four in OCCA tissues were 48% (23/48) and 10% (5/48) respectively (Fig. 1). All five OCCA cell lines were unmethylated in the promoters of both sFRP2 and sFRP4 genes.

Figure 1.

 Determination of sFRP1, sFRP2, sFRP4 and sFRP5 promoter methylation status. (a) The representative gel pictures of promoter methylation of sFRP1 gene and (b) the representative gel pictures of promoter methylation of sFRP5 gene in OCCA (CCC-A∼L), cell lines (OSE2a, OSE2b-2, RMG-I, RMG-II, ES-2, KK, HAC-2), OSA (S1∼S15) and benign endometriotic cysts (B1∼15) analysed by MSP-PCR on sodium bisulfite treated genomic DNA – the lanes corresponding to PCR reactions specific for unmethylated (U) and methylated (M) templates are labelled. Normal lymphocyte DNA supermethylated with SssI methyltransferase, and subsequently treated with sodium bisulfite served as the methylated control (P). Normal lymphocyte DNA treated with sodium bisulfite alone was included as an unmethylated control (Lym). (c) MSP-PCR analysis of promoter methylation of sFRP genes in all tested OCCA samples.

The sFRP5 promoter was significantly methylated in 65% (31/48) of OCCA tissues including frozen and paraffin block samples (Fig. 1), compared with 13·3% (2/15) in OSA and 0% (0/15) in both benign endometriotic cysts and normal ovarian epitheliums (P < 0·0001). There is no statistical difference in the percentage of promoter methylation of sFSRP5 between fresh frozen vs. paraffin-embedded tissues [64% (7/11) vs. 65% (24/37), P > 0·05]. sFRP5 promoter methylation was detected in all the OCCA cell lines tested as well as ovarian epithelium OSE2a and OSE2b-2 cells, implying that sFRP5 promoter methylation may be present in OCCA.

The promoter methylation profile of other tumour suppressor genes in OCCA

To investigate whether other TSGs with aberrant promoter methylation reported in other cancers can also be detected in OCCA, DNA methylation in the CpG islands within the promoters of APC, BRCA1, RB1, p16INK4a, p15INK4b, p14ARF and survivin were analysed with all samples. Results showed that promoter methylation of APC was 28% in OCCA, in contrast to 13·3% in OSA and 0% for benign endometriotic cysts and normal ovarian epithelial tissues (P = 0·097). Methylation at BRCA1 promoter was noted in 33·3% of OSA in contrast to 0% in OCCA and 13·3% in benign ovarian cysts and normal ovarian epithelial tissues (P = 0·006).

High prevalence of promoter methylation in survivin, which encodes an inhibitor of the apoptosis protein, was detected in 72% of OCCA (13/18), 53% of OSA (8/15), and in 93% of benign ovarian cysts and normal ovarian epithelial tissues (14/15) (P = 0·438), indicating that a general defect in survivin may account for tumorigenesis of benign and malignant ovarian tumours. Furthermore, promoters of RB1, p16INK4a, p15INK4b and p14ARF genes were detected unmethylated in pooled or individual samples including OCCA (tissues and cell lines), OSA, benign endometriotic cysts and normal control tissues (data not shown).

Demethylation increases sFRP5 expression in OCCA cells

To verify a role of DNA methylation in the repression of sFRP5 gene expression, selective OCCA cell lines (KK, HAC-2, RMG-I and RMG-II) were treated with the demethylation reagent 5-aza-2-dC for 15 days followed by RT-qPCR analysis. After normalization, relative sFRP5 mRNA level was significantly increased in RMG-II, KK and HAC-2 cells by 5-aza-2-dC (Fig. 2a). However, similar effect of 5-aza-2-dC was not detected in RMG-I cells. Promoter methylation of sFRP5 gene in RMG-II cells was suppressed on day 15, but not on day 3 or day 7, after 5-aza-2-dC treatment (Fig. 2b). Results showed that 5-aza-2-dC significantly decreased the growth of KK cells at 96 h post-treatment (P = 0·031), and decreased the growth of HAC-2 cells at 72 h post-treatment (P = 0·006) (Fig. 2c). In addition, 5-aza-2-dC significantly decreased RMG-I and RMG-II cell growth at 96 h post-treatment (P = 0·003 and P = 0·000 respectively).

Figure 2.

 Demethylating agent (5-aza-2-dC) increased the expression level of sFRP5 in OCCA cell lines. (a) RT-qPCR analysis for sFRP5 gene was performed by preparing RNA from RMG-I, RMG-II, KK and HAC-2 cells with (+Aza) or without (−Aza) treatment. GAPDH was used as a control. *P < 0·01 compared with the untreated control. (b) MS-PCR analysis of methylation status of promoter region of the sFRP5 in RMG-II cells after 3, 10 and 15 day-treatment of 5-aza-2-dC. (c) Growth curve analysis of KK, HAC-2, RMG-I and RMG-II cells upon 5-aza-2-dC treatment. A number of cells were counted using trypan-blue exclusion staining at 48, 72, 96 and 120 h after 5-aza-2-dC treatment. *P < 0·05 compared with the untreated control at indicated time point.

Correlation between sFRP5 methylation and clinicopathological parameters

To confirm the functional consequence of the aberrant promoter methylation of sFRP5 gene, the status of promoter methylation was correlated to the clinicopathological parameters of patients with OCCA (Table 2). With a median follow-up of 44 months, the expected 5-year overall survival (OS) for patients with methylated sFRP5 promoter were significantly worse than that for those with unmethylated sFRP5 (52% vs. 88%, P = 0·03 by log rank test) (Fig. 3). In univariate Cox regression analysis, patients with optimal debulking surgery (P = 0·018), early stage (P = 0·038), the age of < 51 years (P = 0·035) and unmethylated sFRP5 promoter (P = 0·056) showed good prognostic factors in OS. In multivariate analysis, patients with unmethylated sFRP5 promoter (P = 0·017) and optimal debulking (P = 0·032) were independent prognostic factors in OS (Table 3a,b).

Table 2.   Relationship between promoter hypermethylation of sFRP5 and the clinicopathological features of the ovarian clear cell adenocarcinoma examined
NoStageAgeOP statussFRPRecurrenceSurvival
Figure 3.

 The overall survival differences between patients with methylated and unmethylated SFRP5 promoter regions.

Table 3.   (a) Univariate analysis of risk factors for overall survival in patients with OCCA. (b) Multivariate analysis of risk factor for overall survival in patients with OCCA
 Overall survival
 PRelative risk (95% CI)
Surgery (sub optimal vs. optimal)0·0183·85 (1·26–11·74)
FIGO stage (III/IV vs. I/II)0·0383·13 (1·06–9·26)
Age (≥ 50 y/o vs. < 50 y/o)0·0353·49 (1·09–11·20)
sFRP5 (methylated vs. unmethylated)0·0564·32 (0·97–19·36)
CA-125 (> 1000 u/mL vs. < 1000 u/mL)0·162·51 (0·69–9·11)
Surgery (sub optimal vs. optimal)0·0325·29 (1·16–24·24)
FIGO stage (III/IV vs. I/II)0·6691·363 (0·33–5·65)
Age (≥ 50 y/o vs. < 50 y/o)0·0513·30 (0·99–10·95)
sFRP5 (methylated vs. unmethylated)0·0176·65 (1·41–31·43)


To reveal the molecular mechanism of OCCA, we chose and analysed the promoter methylation of some tumorigenesis-related genes, including genes encoding the RB1, p16INK4a, p15INK4b, p14ARF, BRCA1, survivin, APC and Wnt antagonist. This is because that OCCA is more often detected in early stage and its lower proliferation rate compared with OSA. In addition, protein level of survivin was shown to associate with the progression of OCCA [19] and hypermethylation of BRCA1 was detected in serum, plasma and peritoneal fluid of ovarian cancer patients [20].

To the best of our knowledge, this is the first study to demonstrate that a high percentage of DNA methylation in sFRP5 gene promoter detected in OCCA, implicating the importance of sFRP5 in the pathogenesis of OCCA. However, there is still a certain percentage of promoter methylation at sFRP1 (33%), sFRP2 (48%), sFRP4 (10%) and APC (28%) genes present in OCCAs, further suggesting a possible involvement of the Wnt/β-catenin pathway in the molecular tumorigenesis of OCCA. Moreover, the mRNA level of sFRP5 was increased in some OCCA cell lines upon long-term (15 days) 5-aza-2-dC treatment, confirming the relationship between methylation status of promoter and gene expression. The unresponsiveness of 5-aza-2-dC in RMG-II cells indicates that the cell line may tolerate to this treatment as a result of 5-aza-2-dC-induced decrease in cell proliferation which was indeed detected in this study. As most of the tumour samples analysed in this study were paraffin-embedded tissues, it is difficult to evaluate the expression level in parallel with promoter methylation status of sFRP5 in tumour samples. Whether the methylation status of sFRP5 has a direct impact on the expression level of gene in tumour samples will need further evaluation to support the conclusion that the silencing of sFRP5 has a pathological role in OCCA.

Furthermore, our data showed that the overall survival of patients with methylated sFRP5 promoter was significantly worse than those with unmethylated sFRP5 promoter (P = 0·03, by log rank test). After adjusting for age, stage and residual disease after primary surgery, patients with unmethylated sFRP5 promoter still had an independent good prognostic factor in OS in multivariate Cox regression analysis (P = 0·017). Detection of methylation for critical loci may be potentially useful in classification of ovarian cancer and in prognoses of OCCA. We have used real-time qPCR based methylation assay to validate the promotor methylation status of sFRP5 gene in most of the tissue samples, and the results are consistent with those detected by MS-PCR (data not shown). In the future, a more quantitative microarray-based assay should be used to validate the degree of methylation of sFRP5 gene and its clinical correlation with survival of patients with OCCA.

sFRP5 and sFRP2 are inactivated by promoter methylation in colorectal cancer [21]. Besides, sFRP1 is infrequently inactivated by mutation in colorectal cancer [21,22] but is frequently inactivated by promoter methylation in hepatoma [23], bladder cancer [24], mesothelioma [25], ovarian cancer [14,17] and renal cell carcinoma [15]. The 33% (12/36) of sFRP1 promoter methylation detected in this study was higher than the 12% detected by a previous study in 17 ovarian cancer samples including serous, mucinous, endometrioid and clear cell type. The sFRP1 inactivation through promoter methylation may also be involved in pathogenesis of OCCA and explain why some ovarian cancers show increased β-catenin expression in the absence of gene mutations [26].

Taken together, data here suggest that impairment of Wnt antagonists such as sFRP5 caused by promoter hypermethylation may be involved in the pathogenesis of OCCA and may be a potential epigenetic marker for improved prognosis and treatment of OCCA. The role of promoter hypermethylation of sFRP in OCCA needs further confirmation using larger sample sizes.


This work has been supported by a research grant from the National Science Council, Taiwan (93WHK0705006; NSC 94-2314-B-281-005; NSC 96-2320-B-038-028) and research funds from Cathay General Hospital, Taipei, Taiwan (93CGH-TMU-11; 95CGH-TMU-05; 97CGH-TMU-1; 98CGH-TMU-09-2).

Conflict of interest

The authors declare that there are no conflicts of interest.


Gynecologic Cancer Center, Department of Obstetrics and Gynecology, Cathay General Hospital, Taipei, Taiwan (C.-M. Ho, T.-Y. Chien); Department of Pathology, Cathay General Hospital, Taipei, Taiwan (S.-H. Huang); School of Medicine, Fu Jen Catholic University, Hsinchuang, Taipei Hsien, Taiwan (C.-M. Ho); School of Medicine, Taipei Medical University, Taipei, Taiwan (C.-M. Ho); Department of Medical Research, Cathay General Hospital, Sijhih, Taipei Hsien, Taiwan (C.-M. Ho); Department of Obstetrics and Gynecology, Tri-Service General Hospital, Taiwan (H.-C Lai); Department of Pathology, National Taiwan University Hospital, Taipei, Taiwan (M.-C. Lin); Graduate Institute of Medical Sciences, Taipei Medical University, Taipei, Taiwan (S.-F. Chang).