DNA methylation status of REIC/Dkk-3 gene in human malignancies

Purpose The REIC (reduced expression in immortalized cells)/Dkk-3 is down-regulated in various cancers and considered to be a tumor suppressor gene. REIC/Dkk-3 mRNA has two isoforms (type-a,b). REIC type-a mRNA has shown to be a major transcript in various cancer cells, and its promoter activity was much stronger than that of type-b. In this study, we examined the methylation status of REIC/Dkk-3 type-a in a broad range of human malignancies. Methods We examined REIC/Dkk-3 type-a methylation in breast cancers, non-small-cell lung cancers, gastric cancers, colorectal cancers, and malignant pleural mesotheliomas using a quantitative combined bisulfite restriction analysis assay and bisulfate sequencing. REIC/Dkk-3 type-a and type-b expression was examined using reverse transcriptional PCR. The relationships between the methylation and clinicopathological factors were analyzed. Results The rate of REIC/Dkk-3 type-a methylation ranged from 26.2 to 50.0% in the various primary tumors that were examined. REIC/Dkk-3 type-a methylation in breast cancer cells was significantly heavier than that in the other cell lines that we tested. REIC/Dkk-3 type-a methylation was inversely correlated with REIC/Dkk-3 type-a expression. There was a correlation between REIC/Dkk-3 type-a and type-b mRNA expression. REIC/Dkk-3 type-a expression was restored in MDA-MB-231 cells using 5-aza-2′-deoxycytidine treatment. We found that estrogen receptor–positive breast cancers were significantly more common among the methylated group than among the non-methylated group. Conclusions REIC/Dkk-3 type-a methylation was frequently detected in a broad range of cancers and appeared to play a key role in silencing REIC/Dkk-3 type-a expression in these malignancies.


Introduction
Accumulating evidence suggests that tumor progression is governed not only by genetic changes intrinsic to cancer cells but also by epigenetic changes. In cancer epigenetics, aberrant CpG methylation in the promoter region is a key mechanism for gene inactivation, resulting in tumorigenesis in human malignancies (Toyooka and Shimizu 2004).
The REIC (reduced expression in immortalized cells)/ Dkk-3(Dickkopf-3) cDNA, which was expressed in human normal cells and was down-regulated in human immortalized cells and human tumor-derived cells, was identiWed using a representative diVerence analysis system (Tsuji et al. 2000). The amino acid sequence revealed that the REIC gene product was human Dkk-3, one of the Dkk family members. The Dkk family of secreted proteins consists of four members, which share two conserved cysteine-rich domains (Glinka et al. 1998;Krupnik et al. 1999). Dkk-1, the best-characterized member of the Dkk family, functions as a Wnt antagonist or agonist by binding to and inhibiting or activating the Wnt coreceptor LRP6 (BaWco et al. 2001). Unlike Dkk-1, Dkk-2, and Dkk-4, however, REIC/Dkk-3 was recently shown to inhibit TCF-4 receptor activity in lung cancer cells (Yue et al. 2008). TCF-4 activates c-Myc and cyclin D1 through the Wnt/beta-catenin pathway and promotes tumor invasion and metastasis. Because REIC/ Dkk-3 is down-regulated in a variety of malignancies and the overexpression of REIC/Dkk-3 suppresses cell growth, REIC/Dkk-3 has been proposed to act as a tumor suppressor Kurose et al. 2004). Hypermethylation and the down-regulation of REIC/Dkk-3 were observed in a variety of malignancies including non-small-cell lung cancers (NSCLCs) (Kobayashi et al. 2002;Licchesi et al. 2008), gastrointestinal cancers (Maehata et al. 2008), renal clear cell carcinoma (Kurose et al. 2004), acute lymphoblastic leukemia (Roman-Gomez et al. 2004) and osteosarcomas (Hoang et al. 2004). We previously showed the therapeutic eVect of REIC/Dkk-3 in prostate cancers (Abarzua et al. 2005;Edamura et al. 2007) and malignant pleural mesothelioma (MPM) (Kashiwakura et al. 2008). In addition, tumor suppression by REIC/Dkk-3 has also been conWrmed in other malignant tumors (Hsieh et al. 2004;Hoang et al. 2004).
REIC/Dkk-3 mRNA has two isoforms (type-a,b; Gen-Bank accession AB057804). Many papers have described the methylation status in the promoter of REIC/Dkk-3 type-b (Licchesi et al. 2008;Maehata et al. 2008;Veeck et al. 2009). However, the promoter of REIC/Dkk-3 type-a also seems to be important, since Kobayashi et al. (2002) (the group that Wrst identiWed the REIC/Dkk-3 in immortalized cells) have demonstrated that the promoter activity of REIC/Dkk-3 type-a (major promoter) had an approximately 26-fold stronger eVect than that of REIC/Dkk-3 type-b (minor promoter) in a luciferase assay, and the major transcript was REIC/Dkk-3 type-a in various cancer cells they tested. They suggested that hypermethylation of the major promoter (type-a) was a major mechanism for the downregulation of REIC expression. They also suggested that the methylation of the minor promoter (type-b) was accompanied with that of major promoter (type-a) in most cases except four lung cancer cells that they tested. Regardless, those four lung cancer cells had type-b hypermethylation, REIC/Dkk-3 type-b expression was detected in those four lung cancer cells. So they discussed the possibility that minor promoter (type-b) was utilized for the expression in a tissue-speciWc manner, as seen in dual promoter of APC gene.
In this study, we examined the DNA methylation of REIC/Dkk-3 type-a in various kinds of cancers by quantita-tive combined bisulWte restriction analysis (qCOBRA) and investigated the correlation between the REIC/Dkk-3 type-a methylation and REIC/Dkk-3 type-a expression. The qCO-BRA assay can provide more reliable results because the conventional methylation-sensitive restriction enzyme assay that Kobayashi et al. (2002) performed was recently known to be prone to false-positive results due to spurious incomplete digestion (Xiong and Laird 1997). We also analyzed the correlation between REIC/Dkk-3 type-a and typeb expression in various cancer cell lines. Furthermore, we examined the correlation between REIC/Dkk-3 type-a methylation and the clinicopathological features of primary tumors.

Materials and methods
Clinical samples and cell culture Surgically resected specimens of 37 primary breast cancers, 42 primary NSCLCs, 21 primary gastric cancers, 20 primary colon cancers, and 7 MPMs were obtained from Okayama University Hospital (Okayama, Japan), 6 MPMs were obtained from Okayama Rousai Hospital (Okayama, Japan), 5 MPMs were obtained from National Sanyo Hospital (Yamaguchi, Japan), and 27 MPMs were obtained from Karmanos Cancer Center (MI). Ten corresponding non-malignant breast tissues and 10 non-malignant lung tissues were also examined. All tissues were frozen with the liquid nitrogen immediately after surgery and were stored at ¡80°C until extraction of DNA. Institutional Review Board permission and informed consent were obtained for all cases.

DNA extraction and DNA methylation modiWcation
Genomic DNA was extracted from the surgically resected frozen samples and cultured cells by digestion with SDS/ proteinase K followed by phenol/chloroform (1:1) extraction and ethanol precipitation. Two micrograms of each DNA was treated with EZ DNA Methylation Kit (ZYMO RESEARCH, Orange, CA), following the manufacturer's instructions, and was stored at ¡20°C until use.

Quantitative COBRA assay
Nested PCR was carried out using bisulWte-treated DNA followed by the restriction enzyme digestion. First-round touchdown PCR was performed under the following conditions: 95°C for 12 min, 40 cycles of 94°C for 45 s, annealing temperature between 58 and 56°C for 1 min, 72°C for 3 min, followed by Wnal extension step at 72°C for 7 min in a 25-l reaction mixture containing 67 mM Tris-HCl (pH 8.8), 16.6 mM (NH 4 ) 2 SO 4 , 6.7 mm MgCl 2 , 10 mM -mercaptoethanol, 1.25 mM of each deoxynucleotide triphosphate (dNTP) mixture, 0.5 M of each primer, 0.5 unit of HotStar Taq DNA Polymerase (Qiagen, Valencia, CA), and 100 ng of bisulWte-treated DNA. Second-round touchdown PCR was performed using 0.4 l of the Wrst-round PCR products as a template under same condition, but 47 cycles. Universal methylated DNA and universal unmethylated DNA were used for positive control and negative control, respectively. The location of the CpG dinucleotides in the exon1 and in the 5Ј-Xanking region of REIC/Dkk-3 is shown in Fig. 1. Primers were designed using Primer Express software ver.1.0 in the promoter region of REIC/Dkk-3 type-a. Primers for the Wrst-round PCR were REIC-COBRA-F1 5Ј-T GGGTTGTTGTAAGTTTGAAGGT-3Ј and REIC-COBRA-R1 5Ј-CTCACCCACCCCRACTAAAC-3Ј. Primers for the second-round PCR were as follows: REIC-COBRA-F2 5Ј-T GAAGGTTAGATAAGAYGGGTTTAGG-3Ј and REIC-COBRA-R2 5Ј-ACCCACCCCRACTAAACCRAAT-3Ј. These primers were designed to ensure ampliWcation of both methylated and unmethylated forms. Two microliters of second PCR products were digested with 3 units of BstUI (whose restriction site is CGCG) for the restriction fragment length polymorphism analysis. The amplicon of second PCR was named RRCOBRA (Region for REIC-COBRA), and the 5 restriction sites of BstUI are shown in Fig. 1. The digested PCR products were visualized on 3% agarose gels stained with ethidium bromide. The percentages of digested band were analyzed by NIH ImageJ 1.37 V software (http://rsb.info.nih.gov/ij) as described previously (Xiong and Laird 1997). We performed linear regression analysis of qCOBRA with nested PCR using serial dilution to examine whether qCOBRA with nested PCR really reXected % methylation. We diluted unmethylated DNA amplicon with methylated amplicon to make serial dilution (% methylated DNA;0,10,20,30,50,70,80,90,100%) and performed qCOBRA, as described above.
BisulWted DNA sequencing analysis RRCOBRA was cloned into pCR2.1-TOPO Vector using TOPO TA cloning kit (Invitrogen Life Technologies, Carlsbad, CA) following manufacturer's instructions. To determine the methylation status in the promoter lesion of REIC/ Dkk-3 gene, Wve breast cancer cell lines (MCF-7, MDA-MB-231, ZR75-1, HCC1806, and HCC1599) and a lung cancer cell line (H1299) were examined. Seven individual clones from each cell line were sequenced using the dGTP BigDye terminator v3.1 Cycle Sequencing Kit with the ABI PRISM 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA).

RNA extraction and reverse transcriptional (RT)-PCR
Total RNA was extracted from cultured cells using RNeasy Mini Kit (Qiagen, Valencia, CA) following the manufacturer's instruction. Oligo(dT)-primed cDNA was synthesized using Super-Script II (Qiagen, Valencia, CA) with DNase treatment. RT-PCR was carried out in 20 l of reaction mixture with 1xPCR buVer, 200 M of dNTP, 0.3 M of each primer, 0.5 units of HotStarTag DNA Polymerase, and 100 ng of cDNA. A touchdown PCR was performed for REIC/Dkk-3 type-a and type-b under the following conditions: 95°C for 12 min, 35 cycles of 94°C for 30 s, annealing temperature between 62 and 58°C for 1 min, 72°C for 3 min, followed by Wnal extension step at 72°C for 7 min. As an internal control, RT-PCR for GAPDH was carried out under the following conditions: 95°C for 12 min, 35 cycles of 94°C for 45 s, 55°C for 90 s, 72°C for 90 s, followed by Wnal extension step at 72°C for 7 min. The primers for REIC/Dkk-3 type-a expression were REIC (a)-F 5Ј-GGGAGCGAGCAGATCCAGT-3Ј (exon1a) and REIC(a)-R 5Ј-TTTGTCCAGTCTGGTTGTTGGT-3Ј (exon3). The primers for REIC/Dkk-3 type-b expression were REIC(b)-F 5Ј-TGGGAGCTATTAGCGTAGAGGA T-3Ј (exon1b) and REIC(b)-R 5Ј-CATTGTGATAGCTGG GAGGTAAG-3Ј (exon3). The PCR products were visualized on 2% agarose gels stained with ethidium bromide. The bands were analyzed using NIH ImageJ 1.37 V software. The expression ratio in each cell line was deWned as the ratio of particular sample when compared to those of H1299. To conWrm the responsibility of DNA methylation for REIC/Dkk-3 silencing, we treated heavily methylated ER, PgR, and HER2 status in primary breast cancers Estrogen receptor (ER), progesterone receptor (PgR), and HER2 status in primary breast cancers were obtained from patient medical records. HER2 positive was deWned as a score of 2+ and 3+ by immunohistochemistry.

Statistical analyses
The frequencies of REIC/Dkk-3 methylation between two groups were compared using the Fisher's exact test or Mann-Whitney's U test when appropriate. Probability value less than 0.05 was deWned as being statistically sig-niWcant. All data were analyzed by JMP9 for Windows (SAS Institute, Cary, NC).

DNA methylation status in the promoter region of REIC/Dkk-3 type-a
The results of bisulWte genomic DNA sequencing of RRCOBRA are shown in Fig. 1. Each CpG in the 5Ј-Xanking region and in exon1a was heavily methylated in MCF7, MDA-MB-231, and ZR75-1. The CpGs in the 5Ј-Xanking region of exon1a were lightly to moderately methylated, but the CpGs in exon1a were rarely methylated in HCC1806 and HCC1599. In contrast, most of the CpGs were rarely methylated in H1299. We performed the linear regression analysis using the nested qCOBRA and conWrmed the quantitative capacity (data not shown). Representative examples of the COBRA assay in breast cancer cell lines are shown in Fig. 2a. The percentages of REIC/Dkk-3 type-a methylation were calculated by qCOBRA in each cell line and primary tumor (Fig. 2b, c, respectively) and summarized in Table 1. We decided the samples with more than 10% of digested bands as methylation positive in this study. Aberrant methylation was detected in 7 of the 7 (100%) breast cancer cell lines, 16 of the 37 (43.2%) primary breast cancers, 5 of the 11 (45.4%) lung cancer cell lines, 11 of the 42 (26.2%) primary lung cancers, 0 of 4 (0%) MPM cell lines, 7 of the 27 (25.9%) USA primary MPMs, 7 of the 18 (38.9%) Japanese primary MPMs, 8 of the 21 (38.1%) primary gastric cancers, and 10 of the 20 (50.0%) primary colon cancers. REIC/Dkk-3 methylation was not detected in 10 normal breast tissues and 10 normal lung tissues (data not shown). The methylation of REIC/Dkk-3 type-a in the breast cancer cell lines was more frequent than that in the lung, MPM, and prostate cancer cell lines (p = 0.02, p = 0.01, and p = 0.04, respectively). However, no signiWcant diVerences in methylation were observed among the primary breast, lung, MPMs, gastric, and colon cancers (Fig. 2c). The results of qCOBRA in Wve breast cancer cell lines and a lung cancer cell line (H1299) corresponded with the results of bisulWte sequencing.

REIC/Dkk-3 mRNA expression in cell lines and correlation with qCOBRA assay
Representative example of RT-PCR for REIC/Dkk-3 type-a and type-b in the breast cancer cell lines was shown in Fig. 2d. Expression of REIC/Dkk-3 type-a was only detected in HCC1599 and HCC1806 cells, which rarely to moderately harbored REIC/Dkk-3 methylation. Expression of REIC/Dkk-3 type-b was also detected in HCC1599 and HCC1806 cells. The results of expression of REIC/Dkk-3 type-a and type-b in all cell lines (n = 28) were summarized in Table 2. There was a correlation between the expression of REIC/Dkk-3 type-a and type-b (p < 0.01). The relative expressions of REIC/Dkk-3 type-a mRNA and the percentages of REIC/Dkk-3 type-a methylation are shown in Fig. 3. The expression of REIC/Dkk-3 type-a and REIC/Dkk-3 type-a methylation was inversely correlated in the cell lines that were examined (p < 0.01). To conWrm that methylation was responsible for the gene silencing, heavily methylated MDA-MB-231 cells were treated with 5-Aza-CdR. REIC/ Dkk-3 type-a mRNA expression was restored by the treatment of 5-Aza-CdR in a dose-dependent manner (Fig. 4).

REIC/Dkk-3 methylation and clinicopathological correlation
We next examined the relationships between the REIC/ Dkk-3 methylation status and the clinicopathological  Table 3. For the breast cancers, we observed that ER-positive cases were more common in the methylated group than in the non-methylated group (p = 0.03). No signiWcant relationships between REIC/Dkk-3 methylation and the other clinicopathological factors were observed.

Discussion
In this study, we demonstrated that arbitrary CpG methylation in REIC/Dkk-3 type-a promoter region was frequently observed in solid malignancies. Regarding qCOBRA in this study, we conWrmed the accuracy of this assay by linear regression analysis because we performed nested PCR. We decided the samples with more than 10% of digested bands as methylation positive to maintain compatibility with conventional COBRA assay, as we could distinguish 10% of digested band on the agarose gel electrophoresis. Colella et al. (2003) also used a 10% threshold to declare methylation when qCOBRA was compared with pyrosequencing methylation analysis. So a 10% threshold seems to be reasonable criteria to distinguish methylation positive. We examined cell lines using qCOBRA assay, and the accuracy of the qCOBRA was also conWrmed by bisulfate genomic DNA sequencing and linear regression analysis.
The REIC/Dkk-3 type-a methylation, which was determined using qCOBRA assay, revealed to be inversely correlated with the REIC/Dkk-3 type-a mRNA expression among the cell lines (p < 0.01), and the restoration of REIC/ Dkk-3 type-a expression by 5-Aza-CdR treatment was observed in a REIC/Dkk-3 type-a methylated cell line. These results indicate that DNA methylation of REIC/Dkk-3 type-a was responsible for silencing REIC/Dkk-3 type-a expression. As reported previously, we conWrmed that there was a correlation between REIC/Dkk-3 type-a expression and type-b expression in the cell lines that we examined (p < 0.01). Discrepancy of the expression level between REIC/Dkk-3 type-a and type-b was observed in lung cancer cell lines, although the expression of REIC/Dkk-3 type-a completely corresponded with the expression of REIC/Dkk-3 type-b in other cell lines, indicating that REIC/Dkk-3 type-b might be utilized for the expression in a tissue-speciWc manner, as Kobayashi et al. (2002) described.
Among the cancers that were examined, REIC/Dkk-3 type-a methylation was more frequently detected in breast cancer cell lines, although moderate methylation was also observed in other cancers. A previous report showed that the introduction of REIC/Dkk-3 into some breast cancer Prostate cancer 3 of 6 (50%) ND cells had an antitumor eVect (Kawasaki et al. 2009). In addition, the introduction of REIC/Dkk-3 into cancer cells had a direct eVect on the induction of apoptosis and an indirect eVect on the activation of tumor immunity in NK cells through the up-regulation of IL-7 ). Furthermore, REIC/Dkk-3 induces the diVerentiation of human CD14+ monocytes into a novel cell type, resembling immature dendritic cells generated with IL-4 and GM-CSF ). These Wndings support the possible utility of REIC/Dkk-3 gene therapy for not only breast cancers but also a broad range of human malignancies. Indeed, REIC/Dkk-3 gene therapy is ongoing for prostate cancer (http://clinicaltrials.gov/ct2/show/NCT01197209).
Regarding the clinicopathological factors, we found that ER-positive breast cancer was more common in the methylated group than in the non-methylated group in the present study. In contrast, Veeck et al. (2009) reported that there was no correlation between REIC/Dkk-3 methylation and ER and PgR statuses. In other cancers, no sig-niWcant relationship was observed between the REIC/Dkk-3 methylation status and any of the clinicopathological factors that were examined. Previous study has shown that REIC/Dkk-3 methylation was associated with poor survival in primary breast cancers (Veeck et al. 2009). We could not compare these results directly because they examined the REIC/Dkk-3 type-b methylation, which has a lower promoter activity than that of REIC/Dkk-3 type-a. In addition, the number of cases in this study may have been too small for the survival analysis, since the primary purpose of this study was to detect REIC/Dkk-3 type-a methylation and to examine the correlation between methylation and expression. In lung cancer, reduced expression of REIC/Dkk-3 was previously shown to be frequent in poorly diVerentiated adenocarcinoma and squamous cell carcinoma . Further investigations are needed to determine the clinicopathological impact of REIC/Dkk-3 type-a methylation.
In conclusion, we found that the promoter region of REIC/ Dkk-3 type-a was frequently methylated in breast, lung, gastric, colon, and prostate cancers and MPMs. REIC/Dkk-3 type-a methylation and REIC/Dkk-3 type-a mRNA expression were inversely correlated in the cell lines that were examined. Our results suggest that REIC/Dkk-3 type-a methylation is an important mechanism in the pathogenesis of various types of malignancies. Since gene therapy using The expression ratio was deWned as the ratio of particular sample when compared to those of H1299. % methylation was calculated by qCO-BRA assay     REIC/Dkk-3 expressing adenovirus vectors is currently ongoing for the treatment of prostate cancer, similar therapeutic modalities may be applicable for other types of cancers.