Identification of differentially methylated CpG islands in prostate cancer



Epigenetic change such as DNA methylation is one important mechanism for regulating gene expression as genetic change, such as mutation or loss of heterozygosity. Methylation of cancer-related genes has been shown to play an important role in carcinogenesis and tumor progression. Using methylated CpG island amplification (MCA)/representational difference analysis (RDA), we identified four CpG islands in neurotrophin tyrosine kinase receptor type 2 (NTRK2), Protocadherine Flamingo1 and MFPC (Methylated Fragments in Prostate Cancer) 7 and 8. Bisulfite sequencing revealed that 2 regions of NTRK2 as well as MFPC7 and MFPC8 were aberrantly methylated in prostate cancer cell lines, and COBRA showed that 48 (76.24%), 37 (58.7%) and 14 (22.2%) of 63 prostate cancer tissues were methylated, respectively, for these sites. On the other hand, none of 13 benign prostate samples were methylated, except for 1 (7.7%) with NTRK2. For NTRK2, mRNA expression was negative in prostate cancer cell lines (LNCaP and DU145) but was recovered on a methyltransferase inhibitor (5-Aza-CdR) treatment. The role of NTRK2 within NTRK remains unclear. Our results suggest that these 3 hypermethylated DNA fragments also may be markers of prostate cancer detection. © 2004 Wiley-Liss, Inc.

Prostate cancer is not as common in Japanese as in Caucasians or African-Americans, but the number of cases is increasing every year. Genetic changes, such as point mutations, loss of heterozygosity and homozygous deletions in many tumor-suppressor genes, are associated with carcinogenesis.1 Many studies have provided evidence of such genetic alterations in prostate cancer.2 More recently, it has also been reported that inactivation of tumor-suppressor genes by epigenetic change due to DNA methylation could play an important role in carcinogenesis and tumor progression.3 DNA methylation, especially in CpG-rich 5′ regions, inhibits transcription by interfering with initiation or by reducing the binding affinity of sequence-specific transcription factors.4 In prostate cancers, inactivation by aberrant methylation has been reported for many genes, such as RARβ2, GST-P, E-cadherin and RASSF1A.5, 6, 7, 8 Methylation profiling analysis of multiple genes has been proposed for the purpose of evaluating biologic characteristics and identifying useful diagnostic indicators of prognosis.9

As methods for determining genes with expression regulated by DNA methylation, restriction landmark genomic scanning (RLGS), methylated CpG island amplification (MCA)/representational difference analysis (RDA) and differential methylation hybridization (DMH) have been reported.10, 11, 12 MCA/RDA is useful to identify CpG-rich DNA fragments, which are methylated in only the target tissue. Toyota et al.11 identified methylated CpG islands in colon cancer, not only in genes known to be involved in neoplasia but also other examples. To identify cancer-related genes controlled by methylation in prostate cancer, we have applied MCA/RDA to a series of prostate cancers and 2 cell lines.


Cell lines and tissue samples

Prostate cancer cell lines (LNCaP and DU145) were obtained from the American Type Tissue Culture Collection (Rockville, MD). Specimens of 63 primary prostate cancer with known clinicopathologic features (age, stage and Gleason Score) and 13 specimens of benign prostates were obtained by surgery, snap-frozen and stored at −80°C. All benign samples were examined by pathologists and determined to have no precancerous lesion such as high-grade prostatic intraepithelial neoplasia (HGPIN) and no incidental cancer.

Methylated CpG island amplification/representational difference analysis (MCA/RDA)

MCA/RDA was performed as described previously.11, 13 Briefly, 5 μg of DNA was digested with 100 units of SmaI and 20 units of XmaI (New England Biolabs, Beverly, MA) in this order. After these treatments, methylated CpG sites have sticky ends. The restriction fragments were ligated to the RMCA adaptor using T4 DNA ligase (New England Biolabs), and 3 μl of the products were amplified in a 100 μl reaction mixture consisting of 10 mM Tris-HCl (pH 8.3), 1.5 mM MgCl2, 50 mM KCl, 0.5 M betaine, 2% DMSO, 200 μM each dNTP mixture, 100 pmol of RMCA 24 mer primer and 15 units of Taq polymerase (Invitrogen, Carlsbad, CA). The reaction mixture was incubated at 72°C for 5 min and at 95°C for 3 min and then was subjected to 25 cycles of 1 min at 95°C and 3 min at 77°C followed by a final extension of 10 min at 77°C. The MCA amplicon from the prostate cancer cell line DU145 was used as the tester for RDA, and an MCA amplicon generated from a mixture of DNA from the 3 different benign prostate was used as the driver. RDA on this MCA amplicon was performed for 3 rounds using JMCA, NMCA and JMCA adaptors in this order. The RDA products were cloned into pBluescript II plasmid vector (Stratagene, La Jolla, CA), sequenced and identified from the database using BLAST.

Bisulfite modification

Genomic DNA (2 μg) was treated with sodium bisulfite as described previously.14 Briefly, after denaturation in 0.3 M NaOH at 37°C for 15 min, sodium bisulfite and hydroquinone were added to final concentrations of 3.1 M and 0.5 mM, respectively. The reaction was performed at 55°C for 16 hr, and desalting was then conducted using Wizard DNA purification resin (Promega, Madison, WI) according to the manufacturer's instructions. Bisulfite modification was completed by 0.3 M NaOH treatment at 37°C for 15 min. Modified DNA was precipitated with ethanol, washed in 70% ethanol, dried and resuspended in 50 μl of distilled water.

Bisulfite genomic sequencing and combined restriction analysis (COBRA)

The methylation status of the of CpG islands was analyzed by bisulfite genomic sequencing and COBRA as described.14, 15 DNA fragments that satisfied criteria for CpG islands were used.16 The 5′-untranslated region (5′-UTR) of the NTRK2 gene contains 2 CpG islands,17 and both were here analyzed (Fig. 1). The 2 μl bisulfite-modified DNA from 2 prostate cancer cell lines, 3 benign prostate tissues and prostate cancer tissues used for RDA were amplified under the conditions and with the primers listed in Table I. The PCR products were subcloned into TOPO TA cloning vector (Invitrogen) and sequenced using the Big Dye terminator method, about 10 positive clones for each sample. The average levels of methylation at each CpG dinucleotide were graded into 5 categories: 0–19%, 20–39%, 40–59%, 60–79% and 80–100%.

Figure 1.

Maps of MFPCs (methylated fragments in prostate cancer) identified by MCA/RDA. Short vertical lines, CpG dinucleotides; upper arrows, MFPC sites; lower arrows, regions for bisulfite sequences around MFPCs. The 5′-untranslated region of the NTRK2 gene has 2 CpG islands, and MFPC2 was identified as the second CpG island. Transcription and translation initiation sites of NTRK2 are from Stoilov et al.17 The MFPCs 4, 7 and 8 correspond to GeneBank accession nos. AC005923, AC092315 and AP003479, respectively.

Table I. Conditions of the Methylation Analysis
CpG islandsPrimer sequenceAnnealing temperature (°C)CyclesLength (bp)
  1. In the primer sequences, Y = C or T, R = G or A

NTRK2 (region 1)F: 5′AGTGAGATTTYGAGGTTTT3′5240209
NTRK2 (region 2)F: 5′AGAGAGGGAGAGTGTTTTAA3′6140253
Flamingo1F: 5′GGAGAATGAGGTAGTAGGT3′6040345

To examine the methylation status of NTRK2 (region 2), MFPC 7 and MFPC8 in prostate samples, we performed COBRA. By using the same primers and PCR conditions as for bisulfite sequencing, bisulfite-treated DNAs were amplified. The PCR products with NTRK2 and MFPC7 were digested with TaqI (New England Biolabs) at 65°C for 4 hr and with MFPC8 were digested with RsaI at 37°C for 4 hr. The digested fragments were subjected to electrophoresis in 4.0% Seakem GTG agarose gels (BMA, Rockland, ME) and visualized by ethidium bromide staining. When digested bands were confirmed, methylation was concluded.

5-Aza-CdR treatment and RT-PCR

The prostate cancer cell lines were treated with a methyltransferase inhibitor, 5-aza-2′-deoxycytidine (Sigma, St. Louis, MO), for 72 hr before harvesting, alone or with a histone deacetylase inhibitor, trichostatin A (Sigma). Total RNA was extracted using Isogen (Nippon Gene, Tokyo), according to the manufacturer's instructions, and 1μg aliquots were used for generation of cDNA using Superscript reverse transcriptase (Invitrogen). The specific primers applied to detect NTRK2 transcripts were as follows: forward, AGT CCA GAC ACT CAG GAT TTG TAC; reverse, CTC CGT GTG ATT GGT AAC ATG. PCR amplification was performed for 30 cycles at 95°C for 30 sec, 64°C for 30 sec and 72°C for 1 min. GAPDH was used as an internal control. The PCR products were subjected to electrophoresis in 2.0% agarose gels and visualized by ethidium bromide staining.

Immunohistochemical analysis

The antibody against TrkB/NTRK2 protein was polyclonal, raised in rabbit (sc-12, Santa Cruz Biotechnology, Santa Cruz, CA) and used diluted to 2 μg/ml in Tris-HCl buffer (0.05 mol/l, pH 7.5) containing 0.1% BSA, 0.2% FCS and 0.1% Triton X100. Frozen tissue sections (5 μm) were fixed in 100% ice-cold methanol and washed in phosphate-buffered saline (PBS). Endogenous peroxide activity was blocked with 3% hydrogen peroxide in methanol for 10 min and preincubated with 20% normal serum diluted in 1% bovine serum albumin at room temperature for 10 min. They were exposed to primary antibody in moisture chambers at room temperature for 1 hr. Detection took place by the conventional labelled-streptavidin-biotin (LSAB-kit, DAKO, Kyoto, Japan) method according to the manufacturer's instructions. The slides were briefly counterstained with haematoxylin and mounted. For negative controls, serial sections of the same samples were used, omitting the antibody for NTRK2 from the staining protocol and substituting it with a nonimmune serum.

Statistical analysis

Associations between methylation status and clinicopathologic characteristics (age, stage, Gleason score and androgen-dependent state) were analyzed by χ2 test with significance set at the level of p < 0.05. All comparisons were conducted using Survival Tool for StatView (Abacus Concepts, Berkeley, CA).



After MCA coupled with RDA, 100 randomly selected clones were analyzed by DNA sequencing, and 14 clones were revealed to be independent, 10 matched to human genome sequences and 2 clones were Alu-repetitive sequences. Finally, 8 different DNA fragments were identified, which included several known genes (Table II). The average lengths and GC% were 532 bp and 59.3%, respectively. In these DNA fragments, NTRK2, Flamingo1, MFPC7 and MFPC8 satisfied the criteria of CpG island (length > 200 bp, GC content > 50 %, CpG score > 0.6).16

Table II. Methylated DNA Fragments Identified by MCA/RDA
CloneGene/ESTLocationAccession no.LengthGC%CpG/GpCCpG score1
  • 1

    CpG score = [(CpG number) × (length)]/[(C number) × (G number)]16.

MFPC1SCG 10 like protein20q13AF217796633537/380.14
MFPC2NTRK2/Exon 3,4,59q22AF4109025606541/500.70
MFPC4Protocadherin Flamingo13p21AC0059236866882/911.04

Analysis of methylation status in CpG islands

Bisulfite sequencing analysis of 5 CpG islands (including 2 for NTRK2) in prostate cancer cell lines and 3 benign prostate samples were performed (Fig. 2). The 2 5′CpG islands in NTRK2 gene were hypermethylated in prostate cancer cell lines, but were hypomethylated in benign prostate. MFPC7 was also methylated in prostate cancer cell lines and not in benign prostates. MFPC7 was methylated in DU145 and not in LNCaP and benign prostates. Flamingo1 was methylated in cancer cell lines and benign prostates. Thus, we concluded that NTRK2, MFPC7 and MFPC8 might be methylated only in cancers.

Figure 2.

Bisulfite sequencing results for CpG islands in prostate cancer cell lines and benign prostate samples. LNCaP and DU145 are prostate cancer cell lines and B186, B188 and B199 are benign prostate tissues. The numbers show the sites of CpG dinucleotides. The numbers for NTRK2 are from Stoilov et al.,17 and those for Flamingo1, MFPC7 and MFPC8 correspond to GeneBank accession nos. AC005923, AC092315 and AP003479, respectively. PCR products covering each sequencing region were cloned into TA-cloning vector, and 10 positive clones were analyzed for each sample. The levels of methylation at each site were graded into 5 categories (shown at the bottom). The 2 regions of NTRK2 and MFPC7 and MFPC8 were hypermethylated only in prostate cancer cell lines.

The results of COBRA

At first, COBRA was performed for the samples using bisulfite genomic sequencing to confirm that the methylation status revealed by this method reflects that by bisulfite sequencing (Fig. 3). Then we performed COBRA for the postulated cancer-specific methylated DNA fragments in prostate tissues. The frequency of methylation in NTRK2, MFPC7 and MFPC8 was 76.2%, 58.7% and 22.2%, respectively (Table III). Only one sample (7.7%) in benign prostate tissue was methylated for NTRK2. Prostate cancer samples had a significantly greater frequency of methylation of NTRK2 and MFPC7 (p < 0.001) than benign prostates, while having a greater frequency of methylation of MFPC8, but it's not significant. As for clinicopathologic features of primary cancer samples (Table IV), the high GS and recurrent groups had a greater frequency of methylation of MFPC7, but their associations were not significant.

Figure 3.

The results of COBRA for 3 CpG island sites in prostate cancer cell lines, benign prostate tissues and prostate cancer tissues. M shows the methylation patterns after digestion with methylation-sensitive restriction enzymes.

Table III. Results of Cobra in Prostate and Other Cancer Cell Lines
  • *

    p < 0.001;

  • **

    p = 0.059.

Benign prostate131 (7.7%)*0 (0%)*0 (0%)**
Prostate cancer6348 (76.2%)*37 (58.7%)*14 (22.2%)**
Cancer cell lines (prostate)22 (100%)2 (100%)1 (50%)
Table IV. Relationships of Methylation Frequency of MFPCS to Clinicopathologic Factors
  1. AMF, accumulation of methylated fragments.

Gleason score      

Effect of 5-Aza-CdR on NTRK2 expression

To further examine the role of DNA methylation in silencing of NTRK2, we treated LNCaP and DU145, which did not express the gene. After treatment with 5Aza-CdR, expression of NTRK2 was restored in a dose-dependent manner, indicating the role of epigenetic silencing (Fig. 4).

Figure 4.

The expression of NTRK2 gene in prostate cancer cell lines. Prostate cancer cell lines were treated with 5-Aza-CdR for 72 hr or/and 100 ng of trichostatin A (TSA) for the last 24 hr. GAPDH was used as an internal control. NTRK2 expression was recovered on 5-Aza-CdR treatment in a dose-dependent manner.

Immunohistochemical analysis of NTRK2

To examine whether aberrant methylation of 5′-region of NTRK2 is related to loss of expression in clinical cases, immunohistochemical analysis was performed. Twenty prostate cancer samples were examined, which were determined for the methylation status of NTRK2. Of these samples, 5 cases with unmethylated status of NTRK2 exhibited NTRK2 protein expression. However, another 15 cases with methylation of NTRK2 had mixed tumor cell populations with loss and weak to moderate expression levels (Fig. 5). In normal prostate tissue, NTRK2 expression was detected in epithelial cells, although the immunohistochemical reactivity varied.

Figure 5.

Immunohistochemical analysis of NTRK2 in human prostate cancers. (a) Mixed tumor cell populations with loss and moderate expression levels were detected. NTRK2 expression exhibited cytoplasmic staining. (b) Most tumor cells exhibiting no expression and a few tumor cells with weak expression levels were detected.


Using MCA/RDA, we identified 3 DNA fragments whose CpG islands were methylated in prostate cancers. One of the fragments was identical to the 5′-UTR of NTRK2 with 2 CpG islands, both of which were found to be hypermethylated in prostate cancer cell lines. We also confirmed that mRNA expression of NTRK2 in LNCaP and DU145 was recovered with 5-aza-CdR treatment, providing clear evidence of a link between methylation and down-expression in prostate cancer.

NTRK is a tyrosine kinase receptor for the neurotrophins, and there are 3 members (NTRK1, 2 and 3).18NTRK signals regulate neural development and maintenance of neural network.19 The normal prostate epithelium express TrkA but neither TrkB nor TrkC. On the other hand, increased expressions of TrkA and TrkC in prostate cancer have been correlated with progression and metastasis.20 It has been suggested that NTRK plays roles in prostatic tumor cell growth, progression and metastasis.20, 21, 22 However, this is not in line with our findings because DNA fragments that are hypermethylated would be expected to contain tumor-suppressor genes. For instance, in medullary thyroid cancer, NTRK2 expression reduced with tumor progression, while, inversely, that of NTRK3 increased. Moreover, in a cancer cell model, expression of exogenous NTRK2 served to limit tumor growth, suggesting that signaling pathways from NTRK2 may be important to the controlled growth of the medullary thyroid cancer cells.23 McGregor's study suggests that the role of 3 members of NTRK are different from each other, and the expressions of members are changed according to tumor progression. Thus, the actions of 3 NTRK members may differ, and further studies are needed to clarify the role of NTRK2 and its methylation in prostate cancer. Whatever the case, our data clearly indicated decreased expression of NTRK2 due to methylation in prostate cancer cell lines and a high frequency of methylation in primary prostate cancers. In addition, we found no or reduced expression of NTRK2 in primary prostate cancers by immunohistochemistry.

Bisulfite sequencing of CpG islands in MFPC7 and MFPC8 showed hypermethylation in prostate cancer cell lines. COBRA further revealed 58.7% and 22.2% of primary prostate cancer samples to be methylated, respectively, in contrast to benign samples. The significance of the methylation of these unknown DNA fragments is unclear, but our results indicate that MCA/RDA is a powerful method for isolation of cancer-specific methylated fragments.

In conclusion, we isolated 3 CpG-rich DNA fragments found to be hypermethylated in prostate cancer. Detection of methylation of these genes may be the useful biomarkers for early detection of prostate cancer among patients with high risk.