Promoter methylation analysis of DKK2 may be a potential biomarker for early detection of cervical cancer

To compare gene expression differences between healthy donors and patients with cervical cancer, all clinicopathological data related to DKK2 expression profiles (GSE6791, GSE7803, GSE9750, and GSE7410) were carefully selected from the Gene Expression Omnibus (GEO) database, The Cancer Genome Atlas (TCGA) database (https://www.cancer.gov/about-nci/organization/ccg/research/structural-genomics/tcga), and Oncomine websites (https://www.oncomine.org/resource/login.html). GraphPad Prism was used to obtain the scatter diagram. DKK2 methylation expression data was mined in University of ALabama at Birmingham CANcer (UALCAN) databases (http://ualcan.path.uab.edu/) [19, 20]. This study was conducted following the Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK) checklist [21].

Ect1/E6E7 (catalog No. AC39960) normal cervical epithelial cell and HPV-positive HeLa (catalog No. TCHu187), HPV-positive CaSki (catalog No. BNCC338223), HPV-negative HT-3 (catalog No. HTB-32), and HPV-negative C33A (catalog No. BNCC337882) cervical cancer cell lines were screened. Cells were purchased from American Type Culture Collection (ATCC) or BeNa Culture Collection (Beijing, China) and cultured in RPMI-1640 medium (Gibco).

We analyzed a series of cervical specimens with histological diagnoses collected between May 2016 and November 2017 at Liaocheng People's Hospital (Shandong, China). At the time of surgical resection or cervical biopsy, 79 invasive cervical cancer tissues and 63 cervical intraepithelial neoplasia (CIN) tissues were obtained. Of the CIN tissues, 25 were classified as low-grade CIN (low-grade squamous intraepithelial lesions, LSIL) and 38 as high-grade CIN (high-grade squamous intraepithelial lesions, HSIL), and the latter included both CIN2 and CIN3. Normal cervical tissue (29 cases) was retrieved from patients with uterine leiomyomas who underwent a hysterectomy. All samples were stored at −80 °C until analysis. Genotyping was used to detect HPV16 and HPV18 viral DNA [20]. All cases had a confirmed diagnosis, as well as confirmed HPV typing and histopathological type. Exclusion criteria included samples from patients undergoing chemotherapy or radiotherapy, immunocompromised patients, pregnancy, diagnosis of other cancers, uterine cervix operations, and chronic or acute viral infection. Documented written informed consent was provided by all patients for tissue sample collection. The present study was approved by the medical ethics committee of Liaocheng People's Hospital, China (No. 2016022), and was consistent with relevant national regulations and laws, including the People's Republic of China Regulations for the Management of Medical Institutions (promulgated by the Order No. 149 of the State Council on February 26, 1994; and revised in accordance with the Decision of the State Council on Amending Some Administrative Regulations on February 6, 2016), and international medical ethics documents, including the principles of the World Medical Association (WMA) Declaration of Helsinki and its contemporary amendments (2013).

DKK2 mRNA expression was examined by reverse transcription-polymerase chain reaction (RT-PCR) analysis as previously described [18]. Total RNA was extracted from cell lines, CIN, normal cervical tissues, and tumor samples using TRIzol reagent (Invitrogen). DKK2 expression level was calculated relative to that of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in each sample by RT-PCR. Primer pairs used for mRNA expression and the size of PCR products are described in Table 1. The products were separated in 2% agarose gels using a DL1000 marker set (catalog No. 3591A; Takara) marker set.

DNA samples were extracted from tissues and cell lines. Genomic DNA (1 μg) was bisulfite-modified and purified following the guidelines described by the manufacturer. After desulfonation with NaOH and precipitation with ethanol, the final products were dissolved in 20 μL of Tris-EDTA buffer and stored at −80 °C.

DKK2 gene methylation was determined using a Methylation-Gold Kit (Zymo Research). Primer pairs used for methylation analysis and PCR size products are shown in Table 1, as previously described [12, 23]. Briefly, 5 min predenaturation occurred at 95 °C, 30 s denaturation for 35 cycles, 30 s annealing at 54 °C (unmethylated-specific PCR) or 58 °C methylation-specific PCR (MSP), and 10 min extension at 72 °C. As a positive control for methylation, DNA obtained from normal tissue was methylated in vitro with M.Sssl CpG methyltransferase (New England BioLabs). A water blank (no template DNA) was also included as a negative PCR control. The products were separated in 2% agarose gels using a DL1000 marker set (catalog No. 3591A; Takara) marker set.

HeLa, CaSki, HT-3, and C33A cells (2 × 105) were treated with 10 μmol 5-Aza-2′-deoxycytidine (5-Aza-dC) (Sigma) for 96 h. After adding the demethylation treatment, the medium was changed every 24 h. RNA and DNA were isolated according to routine procedures.

SPSS Statistics for Windows (version 17.0) was used for all statistical analyses. Student t tests were performed for data mining, and χ² or Fisher exact tests were performed for RT-PCR and MSP analyses. P < 0.05 was considered significant.

We examined DKK2 mRNA expression in 79 normal cervical tissues (50 by data mining, 29 by RT-PCR), 9 cervical cancer lines (HT-3, C4-I, CaSki, MS751, C33A, SiHa, SW756, ME-180, and HeLa cell lines by data mining, HeLa, HT-3, C33A, and CaSki cell lines by RT-PCR), and 498 cervical cancer samples (419 by data mining, 79 by RT-PCR). Based on data from the GEO database (Figures 1A–D), DKK2 mRNA expression was lower in cervical cancer tissues compared with normal tissues (P < 0.0001 for Figures 1A, B, and D, P = 0.0004 for Figure 1C). However, based on data from the TCGA database (Figure 1E), the difference was not significant (P = 0.848), although the levels of DKK2 mRNA expression showed a similar trend to that of the GEO database. Compared with normal cervical epithelial cells, expression in cervical cancer cells including HT-3, C4-I, CaSki, MS751, C33A, SiHa, SW756, ME-180, and HeLa cervical cancer cell lines was lower, as shown in Figure 1F (P = 0.023).

Figure 1

We then examined DKK2 mRNA expression both in cervical cancer cell lines and in tissue samples. Our RT-PCR analysis indicated that DKK2 mRNA expression was weak in HPV-negative C33A cells, and silenced in HPV-negative HT-3 cells and HPV-positive CaSki and HeLa cells, as compared with the normal cervical epithelial cell line Ect1/E6E7 (Figure 1G). All 29 normal cervical tissues, and 46 of 63 (73.0%) CIN samples, showed DKK2 mRNA expression (Table 2). Among 79 cervical cancer samples analyzed, only 25 of 79 (31.6%) showed DKK2 mRNA expression. The positive rate of DKK2 mRNA expression in cervical cancer decreased significantly compared to that in HSIL (χ² = 5.999, P = 0.014). The positive rate of DKK2 mRNA expression in HSIL was significantly lower than that in normal cervical samples (χ² = 17.385, P = 0.00003), and LSIL (χ² = 15.318, P = 0.001). The RT-PCR results for the 2 normal cervical tissues (N1, N2), 2 LSIL (L1, L2), 2 HSIL (H1, H2), and 11 primary tumors (T1 to T11) analyzed are shown in Figure 1H.

We then examined the status of DKK2 promoter methylation in HeLa, HT-3, C33A, and CaSki cell lines by MSP. DKK2 hypermethylation was detected in all 4 cell lines (Figure 2A). Of these 4 cell lines, the C33A cell line showed the weakest DKK2 mRNA expression, although both unmethylated and methylated bands were detected. Correlation analysis between DKK2 silencing and promoter hypermethylation was performed by treating cell lines with 10 μM 5-Aza-dC induction for 4 d, followed by promoter methylation level analysis. DNA-demethylating agent treatment significantly led to DKK2 demethylation and re-expression of the transcript among all 4 cell lines analyzed (Figure 2B).

Figure 2

We then examined DKK2 methylation status in cervical tissues. Methylation levels were significantly higher in cervical cancer tissues, as shown in TCGA samples (Figure 2C, P < 0.0001). MSP analysis was then performed to confirm the overall frequency of DKK2 methylation in cervical cancer. DKK2 promoter hypermethylation was detected in 41 of 79 cases (51.9%) of cervical cancer tissues. However, methylation status was only found in 10 of 63 CIN cases (12.7%), and no methylation status was detected in normal cervical cases (Figure 2D). The positive rate of DKK2 methylation in cervical cancer was significantly higher than that in HSIL (χ² = 8.346, P = 0.004), whereas the positive rate of DKK2 methylation in HSIL was significantly higher than that in normal cervical samples (χ² = 7.934, P = 0.005) and in LSIL samples (χ² = 4.375, P = 0.037) (Table 2). Positive DKK2 mRNA expression in cervical cancer cases without methylation (47.4%) was significantly higher than that of cervical cancer tissues with promoter hypermethylation (17.1%) (χ² = 8.368, P = 0.002).

The correlation between DKK2 hypermethylation and the primary clinicopathological characteristics of cervical cancer, including histology, stage, lymph node metastasis, HPV infection, and differentiation, was investigated further. DKK2 methylation (76.5% vs 45.2%, respectively, χ² = 5.239, P = 0.022) and DKK2 mRNA expression (11.8% vs 37.1%, respectively, χ² = 3.958, P = 0.047) are significantly different between patients with or without lymph node metastasis (Table 3). The rate of detection of DKK2 methylation in HPV16/18-positive patients in the cervical cancer group (66.7%) was significantly higher than that in HPV16/18-negative patients (36.0%) (χ² = 6.279, P = 0.015). The rate of detection of DKK2 methylation in HPV16/18-positive patients (24.1%) in the precancerous lesion group was slightly higher than that in HPV16/18-negative patients (13.0%) (χ² = 1.016, P = 0.482) (Table 4). No significant association was detected between DKK2 hypermethylation and other cervical cancer clinicopathological characteristics, such as clinical stage (P = 0.32), histology (P = 0.55), and tumor grade (P = 0.17) (Table 3).