DNA demethylation of vascular endothelial growth factor-C is associated with gene expression and its possible involvement of lymphangiogenesis in gastric cancer
Version of Record online: 17 JAN 2007
Copyright © 2007 Wiley-Liss, Inc.
International Journal of Cancer
Volume 120, Issue 8, pages 1689–1695, 15 April 2007
How to Cite
Matsumura, S., Oue, N., Mitani, Y., Kitadai, Y. and Yasui, W. (2007), DNA demethylation of vascular endothelial growth factor-C is associated with gene expression and its possible involvement of lymphangiogenesis in gastric cancer. Int. J. Cancer, 120: 1689–1695. doi: 10.1002/ijc.22433
- Issue online: 21 FEB 2007
- Version of Record online: 17 JAN 2007
- Manuscript Accepted: 29 SEP 2006
- Manuscript Received: 5 MAY 2006
- Ministry of Education, Culture, Science, Sports, and Technology of Japan (Grants-in-Aid for Cancer Research)
- Ministry of Health, Labor
- Welfare of Japan
- DNA methylation;
- DNA demethylation;
- gastric cancer
Previous studies have indicated that lymphangiogenesis in solid tumors is associated with lymphatic metastasis. Overexpression of Vascular endothelial growth factor (VEGF)-C plays a major role in lymphangiogenesis in cancers. In the present study, DNA methylation and expression of the VEGF-C gene was investigated in gastric cancer (GC). Four GC cell lines (MKN-45, MKN-74, HSC-39 and HSC-43) showed no expression of VEGF-C, and the VEGF-C gene was found to be methylated in these cells. In contrast, 7 GC cell lines (MKN-1, MKN-7, MKN-28, TMK-1, KATO-III, SH101-P4 and HSC-44PE) expressed VEGF-C, and the VEGF-C gene was found to be unmethylated in these cell lines. In addition, expression of VEGF-C mRNA was retrieved by treatment with a demethylating agent, Aza-2′-deoxycytidine. In GC tissue samples, bisulfite DNA sequencing analysis revealed that VEGF-C was not methylated in 9 (29.0%) of 31 GC samples, whereas demethylation was not observed in corresponding non-neoplastic mucosa samples. Overexpression of VEGF-C mRNA was observed in 16 (51.6%) of 31 GC samples by quantitative reverse transcription-polymerase chain reaction. Of the 9 GC cases with VEGF-C demethylation, 8 (88.9%) overexpressed VEGF-C. In contrast, of the 22 GC cases without VEGF-C demethylation, 8 (36.4%) overexpressed VEGF-C (p = 0.0155). Furthermore, lymphatic vessel density determined by immunostaining of podoplanin in GC tissues was associated with overexpression of VEGF-C (p < 0.0001). These results suggest that demethylation and activation of the VEGF-C gene is likely involved in lymphangiogenesis in GC. © 2007 Wiley-Liss, Inc.
According to the World Health Organization, gastric cancer (GC) is the fourth most common malignancy in the world, with ∼870,000 new cases every year, and mortality from GC is second only to that from lung cancer.1 Despite improvements in diagnostic and therapeutic methods, the prognosis of advanced GC with extensive invasion and metastasis remains poor. Several molecules associated with invasion and metastasis have been identified2, 3; however, all the mechanisms underlying metastasis remains unclear.
We previously reported that hypoacetylation of histone H4 is associated with tumor progression and lymph node metastasis.4 Genes with expression regulated by histone acetylation may be involved in tumor progression or metastasis. In GC, expression of p21WAF1/CIP15 and PINX16 are regulated by histone acetylation, but expression of these genes is not associated with tumor progression or metastasis. Histone deacetylation also plays an important role in CpG island methylation-associated gene inactivation.7 DNA methylation of CpG islands is detected commonly in human cancers including GC.8, 9, 10, 11 Hypermethylation of CpG islands is associated with silencing of several genes,12, 13 especially defective tumor-related genes, and has been proposed as an alternative way to inactivate tumor-related genes in human cancers.14, 15
Several genes whose expression is activated by DNA demethylation have been reported. Demethylation of both MAGE16 and synuclein γ17 are correlated with tumor progression and lymph node metastasis in GC. Activation of matrix metalloproteinase genes by DNA demethylation has been observed in pancreatic cancer cell lines.18 Taken together, the currently available data suggest that certain genes activated by DNA demethylation may be involved in tumor progression and lymph node metastasis.
It has been shown a close association between vascular endothelial growth factor (VEGF) family members and tumor metastasis.19 VEGF has been established as a primary angiogenic molecule involved in development, adult physiology and pathology. VEGF-C and VEGF-D are primarily lymphangiogenic factors, but they can also induce angiogenesis under some conditions. Overexpression of VEGF-C has been detected in a variety of cancers.20 However, the role of demethylation of VEGF-C in human cancer has not been examined.
In the present study, we examined whether DNA demethylation may be associated with the overexpression of the VEGF-C in GC. We show that the overexpression of VEGF-C is associated with DNA demethylation and can be restored in GC cell lines after aza-2′-deoxycytidine (Aza-dC)-induced demethylation. We further investigated DNA demethylation of VEGF-C and its possible involvement in lymphangiogenesis by immunostaining of podoplanin, a marker of lymphatic endothelial cell, in GC.
Material and methods
GC cell lines and drug treatment
Eleven cell lines derived from human GC were used. The TMK-1 cell line was established in our laboratory from a poorly differentiated adenocarcinoma.21 Five gastric carcinoma cell lines of the MKN series (MKN-1, adenosquamous cell carcinoma; MKN-7; MKN-28; MKN-74, well differentiated adenocarcinoma; and MKN-45, poorly differentiated adenocarcinoma) were kindly provided by Dr. Toshimitsu Suzuki. KATO-III cell lines were kindly provided by Dr. Morimasa Sekiguchi. SH101-P4, HSC-39, HSC-43 and HSC-44PE cell lines were kindly provided by Dr. Kazuyoshi Yanagihara.22, 23, 24, 25 All cell lines were maintained in RPMI 1640 (Nissui Pharmaceutical, Tokyo, Japan) containing 10% fetal bovine serum (BioWhittaker, Walkersville, MD) in a humidified atmosphere of 5% CO2 and 95% air at 37°C. To analyze transcriptional activation of VEGF-C, MKN-1, MKN-45 and MKN-74 cells were incubated for 5 days with 1 μM Aza-dC (Sigma Chemical, St. Louis, MO, USA) or for 24 h with 300 nM TSA (Wako, Tokyo, Japan).
Frozen tissue samples were collected from 31 patients (age range, 41–86 years; mean, 67.8 years) with GC who underwent surgery between 1998 and 2001 at the Department of Surgical Oncology, Hiroshima University Hospital (Hiroshima, Japan). All patients underwent curative resection, and all GC samples were advanced GC. These 31 GC tissue specimens and 5 corresponding non-neoplastic mucosa samples from the 5 GC patients (age range, 57–75 years; mean, 68.3 years) were analyzed for methylation of the VEGF-C gene. Total RNA was available for 31 pairs of tumor and corresponding non-neoplastic mucosa. GC and corresponding non-neoplastic mucosa were removed surgically, frozen immediately in liquid nitrogen and stored at −80°C until use. We confirmed microscopically that the tumor specimens consisted mainly of carcinoma tissue (>50%, on a nuclear basis) and that specimens of non-neoplastic mucosa did not show tumor cell invasion or significant inflammatory involvement. Tumor staging was carried out according to the TNM staging system.26 We also examined levels of VEGF-C mRNA in 10 samples of normal gastric mucosa obtained endoscopically from 10 healthy young individuals (age range, 22–35 years; mean, 26.4 years) and evaluated methylation status of the VEGF-C gene in 2 samples from normal healthy young individuals. These healthy volunteers were confirmed to be free of malignancy by gastrointestinal endoscopy and biopsy. Because written informed consent was not obtained, for strict privacy protection, identifying information for all samples was removed before analysis; the procedure was in accordance with the Ethical Guidelines for Human Genome/Gene Research enacted by the Japanese Government.
Conventional and quantitative reverse transcription-polymerase chain reaction analyses
Total RNA was extracted with an RNeasy Mini Kit (Qiagen, Valencia, CA), and 1 μg of total RNA was converted to cDNA with a First Strand cDNA Synthesis Kit (Amersham Biosciences, Piscataway, NJ). Conventional RT-PCR was performed to investigate VEGF, VEGF-B, VEGF-C and VEGF-D expression in GC cell lines. Amplification products were then separated by 1% agarose gel electrophoresis, stained with ethidium bromide, and visualized under UV light. ACTB-specific PCR products served as internal controls. Primer sequences and additional PCR conditions are available upon request.
Quantitation of VEGF-C mRNA levels in human tissue samples was done by real-time fluorescence detection as described previously.27 Primer sequences and annealing temperatures are shown in Table I. PCR was performed with a TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA, USA). Real-time detection of the emission intensity of fluorescent reporter dye was performed with an ABI PRISM 7700 Sequence Detection System (Applied Biosystems) as described previously.28ACTB-specific PCR products were amplified from the same RNA samples and served as internal controls. We calculated the ratio of VEGF-C mRNA levels in GC tissue (T) to those in corresponding non-neoplastic mucosa (N). T/N ratios >2-fold were considered to represent overexpression.
|Primer sequence||Annealing temperature (°C)||Size (bp)|
|Bisulfite DNA sequencing (Region 1)|
|Bisulfite DNA sequencing (Region 2)|
|Conventional RT-PCR (VEGF-C)|
|Conventional RT-PCR (ACTB)|
|Quantitative RT-PCR (VEGF-C)|
|Probe: 5′-FAM CAGCAACACTACCACAGTGTCAGGCA TAMRA-3′|
|Quantitative RT-PCR (ACTB)|
Genomic DNA extraction and bisulfite genomic DNA sequencing
Genomic DNAs were extracted with a Genomic DNA Purification Kit (Promega, Madison, WI). To examine DNA methylation patterns, genomic DNA was treated with 3 M sodium bisulfite as described previously.29 For analysis of DNA methylation of VEGF-C, we performed bisulfite genomic DNA sequencing analysis. Two sets of primers were used to assess the different regions (Regions 1 and 2) of the VEGF-C gene (Fig. 1c). Except for primer complementary sequences, Region 1 contains 18 CpG sites, and Region 2 contains 20 CpG sites. Two-microliter aliquots were used as templates for PCR reactions. Primer sequences and annealing temperatures are shown in Table I. Each target sequence was amplified in a 50-μl reaction containing 0.2 μM dNTPs, 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 2 mM MgCl2, 0.3 μM each primer and 0.75 units of AmpliTaq Gold (Applied Biosystems). PCR amplification consisted of 35 cycles after the initial AmpliTaq Gold activation step.
PCR products were purified and cloned into pCR2.1 (Invitrogen, Carlsbad, CA). The cloned PCR fragments obtained from each sample were sequenced with M13 forward primer and a PRISM AmpliTaq DNA Polymerase FS Ready Reaction Dye Terminator Sequencing Kit (Applied Biosystems). Reamplified DNA fragments were purified with Centri-Sep Columns (Applied Biosystems) and sequenced with an ABI PRISM 310 Genetic Analyzer (Applied Biosystems).
Immunostaining of lymphatic vessels and determination of lymphatic vessel density
Consecutive 4-μm sections of formalin-fixed, paraffin-embedded tissue were cut onto glass slides from each study block. Sections were immunostained for podoplanin. Podoplanin, a 38-kDa membrane glycoprotein originally identified on podocytes, is expressed on the endothelium of lymphatic capillaries but not in quiescent or proliferating blood vascular endothelium.30 Immunostaining was done with the Histofine Simplestain MAX-PO (MULTI) (Nichirei Biosciences, Tokyo, Japan) immunoperoxidase technique. Primary antibody was a mouse antipodoplanin monoclonal antibody (1:200, AngioBio, Del Mar, CA) and was incubated on the sections for 3 h at room temperature. Negative controls were done with nonspecific IgG as the primary antibody. Sections were counterstained with hematoxylin. LVD was evaluated by 2 independent investigators (S.M., N.O.), who were blind to the clinical course of the patients and the VEGF-C expression status of the tumors. In brief, after scanning an immunostained section at low magnification (×100), the area of tissue with the greatest number of distinctly highlighted lymphatic vessels (“hot spot”) at the border of invasive cancer or inside the tumor was selected. LVD was then determined by counting all antipodoplanin immunostained lymphatic vessels at ×200 in an examination area. Only vessels with typical morphology (lumen) were considered lymphatic microvessels. After the 6 areas of highest neovascularization were identified, lymphatic vessels were counted and the average count was determined.
Differences were analyzed statistically by Fisher's exact and Mann–Whitney U tests. p values less than 0.05 were considered statistically significant.
Expression of VEGF, VEGF-B, VEGF-C, and VEGF-D mRNAs in GC cell lines
To examine expression of VEGF genes, we performed conventional RT-PCR analysis of 11 GC cell lines. Expression of VEGF, VEGF-B and VEGF-D was detected in all GC cell lines (data not shown). Four cell lines, MKN-45, MKN-74, HSC-39, and HSC-43, showed no expression of VEGF-C (Fig. 1a). We hypothesized that loss of VEGF-C expression might be caused by DNA methylation or histone hypoacetylation. To test this hypothesis, MKN-1, MKN-45 and MKN-74 cells were treated with Aza-dC or TSA and then subjected to RT-PCR analysis (Fig. 1b). Treatment with Aza-dC induced VEGF-C expression in MKN-45 and MKN-74 cells, whereas treatment with TSA did not. Treatment with Aza-dC or TSA did not significantly change VEGF-C expression in MKN-1 cells. These results suggest that DNA methylation may suppress VEGF-C expression.
Analysis of VEGF-C methylation in GC cell lines
To evaluate the extent of VEGF-C methylation, we performed bisulfite DNA sequencing of genomic DNAs from 7 VEGF-C-positive (MKN-1, MKN-7, MKN-28, TMK-1, KATO-III, SH101-P4 and HSC-44PE) and 4 VEGF-C-negative (MKN-45, MKN-74, HSC-39 and HSC-43) cell lines. Results of bisulfite genomic DNA sequencing are shown in Figure 1c. In VEGF-C-positive cells, all 4 CpG sites (sites 1, 2, 3 and 4) of the 3′ region of VEGF-C were unmethylated in 6 of 10 MKN-1 clones, all MKN-7 clones, all MKN-28 clones, all TMK-1 clones, 7 of 10 KATO-III clones, 4 of 10 SH101-P4 clones and 5 of 10 HSC-44PE clones. In contrast, no VEGF-C-negative cell clones showed unmethylation in all 4 CpG sites in the 3′ region of VEGF-C. In addition, in Aza-dC-treated MKN-45 and MKN-74 cells, all 4 CpG sites in the 3′ region were demethylated. Thus, DNA methylation of the 4 CpG sites plays an important role in transcriptional inactivation of VEGF-C gene, at least in the MKN-45 and MKN-74 cell lines.
DNA methylation status of VEGF-C gene in GC tissue samples
Because overexpression of VEGF-C has been reported in GC,31, 32 we examined whether DNA demethylation is involved in overexpression of VEGF-C in GC tissue samples. Bisulfite DNA sequencing was performed on genomic DNAs from 31 GC samples. Because bisulfite sequencing analyses of the GC cell lines revealed that DNA methylation of 4 CpG sites in the 3′ region (sites 1, 2, 3 and 4) was associated with VEGF-C expression, we analyzed the DNA methylation status of Region 1. Five clones from each GC sample were sequenced. Representative results of bisulfite genomic DNA sequencing analysis are shown in Figure 2a. In general, many CpG sites were methylated; however, the 4 CpG sites of the 3′ region of Region 1 were unmethylated in several GC samples. In most cases, the CpG sites of the 5′ region of Region 1 were densely methylated. On the basis of the data obtained from the GC cell lines, it was considered “unmethylated clone” if all 4 CpG sites 1, 2, 3 and 4 were unmethylated. We regarded the VEGF-C methylation status of a case as “unmethylated” if that case contains at least 1 unmethylated clone. VEGF-C was unmethylated in 9 (29.0%) of 31 GC samples. No association was detected between methylation status of VEGF-C and T grade (p = 0.7043), N grade (p = 1.0000) or tumor stage(p = 0.4564, Table II).
|VEGF-C methylation status||p Value1|
|Stage I/II||6 (35.3%)||11||0.4564|
|Stage III/IV||3 (21.4%)||11|
We then examined DNA methylation status of VEGF-C in 5 samples of corresponding non-neoplastic mucosa. In contrast to the variations in DNA methylation patterns observed in the GC samples, the DNA methylation pattern in non-neoplastic gastric mucosa samples was fairly consistent. With the exception of 2 CpG sites (sites 5 and 9), 5′ and 3′ CpG sites were densely methylated. Importantly, in contrast to GC samples, all 4 CpG sites (sites 1, 2, 3 and 4) were methylated in non-neoplastic samples, suggesting that demethylation of these 4 CpG sites is a cancer-specific event.
Relation between VEGF-C DNA methylation status, mRNA expression and LVD
We measured levels of VEGF-C mRNA by quantitative RT-PCR to investigate whether methylation of VEGF-C was associated with gene expression. Overexpression of VEGF-C mRNA (T/N > 2) was observed in 16 (51.6%) of 31 GC samples (Fig. 2b). Of 9 GC cases with unmethylated VEGF-C, 8 (88.9%) showed overexpression of VEGF-C. In contrast, only 8 of 22 (36.4%) GC cases with methylated VEGF-C showed overexpression (p = 0.0155, Fisher's exact test, Table III).
|VEGF-C overexpression||p Value1|
To investigate whether an association exists between LVD and methylation status or mRNA expression of VEGF-C, we evaluated LVD of 31 GC cases by immunostaining for podoplanin (Fig. 2c). Most podoplanin-positive lymphatic vessels were present at the tumor margin, whereas intratumoral lymphatics were rare as previously reported in other tumors.33, 34 LVD was significantly higher in GC cases with overexpressing VEGF-C than in those without overexpression (p < 0.0001, Fig. 2d). No association was noted between DNA methylation status of VEGF-C and LVD (p = 0.0784, Fig. 2d).
If lack of methylation of VEGF-C is cancer-specific and associated with gene expression, non-neoplastic tissues should not express VEGF-C mRNA. To investigate this hypothesis, we performed quantitative RT-PCR of 10 normal gastric mucosa samples from healthy young individuals. Expression of VEGF-C was not detected in 10 normal gastric mucosa samples from healthy young individuals (Fig. 3a). Bisulfite genomic DNA sequencing was performed for 2 of these 10 samples, and we confirmed that all the 4 CpG sites (sites 1, 2, 3 and 4) were methylated (Fig. 3b). In contrast, expression of VEGF-C was detected in several corresponding non-neoplastic gastric mucosa samples (Fig. 3a). However, the VEGF-C mRNA levels in corresponding non-neoplastic gastric mucosa samples were very low compared with those in GC tissues with VEGF-C unmethylation. As shown in Figure 2a, in samples 1N, 2N, 3N, 4N and 5N, VEGF-C was methylated, but expression of VEGF-C was detected in samples 1N, 2N, 3N and 5N.
VEGF-C is a ligand for VEGF receptor-3, which is expressed on endothelial cells of lymphatic vessels.35 Expression of VEGF-C is associated with the development of lymphatic vessels. The prognostic value of VEGF-C overexpression in GC has been reported; overexpression of VEGF-C is associated with lymph node metastasis and poor prognosis.31, 36 However, the mechanism that underlies overexpression of VEGF-C in cancers remains unclear. In the present study, we demonstrated that demethylation of the VEGF-C gene is associated with expression in GC cell lines and that demethylation of VEGF-C by Aza-dC can activate expression of VEGF-C mRNA although it has been suggested that genes activated by Aza-dC may not result from the direct inhibition of DNA methylation.37 Our results suggest that demethylation of the VEGF-C gene plays an important role in transcriptional activation of VEGF-C in GC. In addition, we found that the VEGF-C gene is frequently unmethylated in GC tissues and that this lack of methylation is associated with overexpression of VEGF-C. We confirmed that VEGF-C is methylated in non-neoplastic gastric mucosa samples from patients with GC, indicating that the lack of VEGF-C methylation in GC samples is due to demethylation of the gene. It is important to note that the source of VEGF-C in GC tissues can be from the GC cells themselves or from stromal cells such as tumor-associated macrophages (TAMs), because in squamous carcinoma of the uterine cervix, a subfraction of TAMs are a major source of VEGF-C.38 Our findings cannot determine in which cell population changes in demethylation occur. Our previous immunohistochemical study has indicated that VEGF-C is expressed in GC cells but not in stromal cells.39, 40 Therefore, we presume that demethylation of the VEGF-C gene might occur in GC cells.
In the present study, several GC samples showed overexpression of VEGF-C mRNA in the absence of DNA demethylation and some GC cases showed partial methylation in the 4 CpG sites (sites 1, 2, 3 and 4). In normal gastric mucosa samples from healthy young individuals, no expression of VEGF-C was observed, but in some non-neoplastic gastric mucosa samples from patients with GC, slight expression of VEGF-C was observed without DNA demethylation. These findings indicate that DNA methylation of the region we analyzed does not completely inactivate VEGF-C expression in some conditions. Alternative activating pathways, such as alteration of transcription factors, may account for the overexpression of VEGF-C in these samples. Although demethylation of VEGF-C by Aza-dC treatment activated expression of VEGF-C mRNA, our findings cannot rule out that Aza-dC treatment indirectly affects VEGF-C expression, for example by demethylation of a transcription factor gene required for VEGF-C expression. Recent studies suggest that Foxc2 can regulate VEGF-C expression (reviewed in Ref.41). It is known that only methylation of a small region within a promoter CpG islands can repress gene transcription.42 Methylation of Exon 1 or a far upstream region can be associated with loss of transcription, but usually does not have a causal role in transcriptional repression. Because slight expression of VEGF-C was observed without DNA demethylation in non-neoplastic gastric mucosa samples from patients with GC, DNA methylation of the region we analyzed may not have a causal role in transcriptional repression. At least however, our present data indicate that DNA demethylation is important for VEGF-C overexpression because most GC samples (88.9%) showing DNA demethylation overexpressed VEGF-C mRNA.
In the present study, high LVD was associated with overexpression of VEGF-C but not DNA demethylation, suggesting that overexpression of VEGF-C caused by DNA demethylation participate partly in lymphangiogenesis in GC. In addition to DNA demethylation, another mechanism may be involved in lymphangiogenesis. Furthermore, there was no association between DNA methylation status of VEGF-C and clinical features, such as lymph node metastasis. Many studies have indicated that VEGF-C levels in primary tumors are correlated with lymph node metastasis in thyroid, prostate, gastric, colorectal, lung and esophageal cancers.43 To produce a metastasis, tumor cells must complete a multistep progression through a series of sequential and selective events,44 and several molecules associated with metastasis have been reported. Therefore, other molecules may affect lymph node metastasis in the present study.
In conclusion, our data clarify one of the mechanisms involved in overexpression of VEGF-C in human GC. Clinical trials of DNA methylation inhibitors as cancer therapeutics are underway.45, 46 Although we did not investigate the potential effect of DNA methylation inhibitors on tumor lymphangiogenesis in vivo, DNA methylation inhibitors could stimulate the tumor cell metastasis by activation of VEGF-C gene expression at least in GC with DNA methylation of VEGF-C. Our data suggest that combinations of DNA methylation inhibitors and VEGF-C inhibitors47 may be more effective anticancer therapeutics.
We thank Mr. Masayoshi Takatani and Mr. Masayuki Ikeda for excellent technical assistance and advice. This work was carried out with the kind cooperation of the Research Center for Molecular Medicine, Faculty of Medicine, Hiroshima University. We thank the Analysis Center of Life Science, Hiroshima University for the use of their facilities.
- 1Stomach cancer. In: Stewart BW, Kleihues P, eds. World cancer report. Lyon: IARC Press, 2003. 197., .
- 26Sobin LH, Wittekind CH, eds. TNM classification of malignant tumors, 6th edn. New York: Wiley-Liss, 2002. 65–8.