*Published 2011. This article is a US Government work and, as such, is in the public domain of the United States of America.
Nonsteroidal anti-inflammatory drug-activated gene, NAG-1, a transforming growth factor-β member, is involved in tumor progression and development. The association between NAG-1 expression and development and progression of glioma has not been well defined. Glioblastoma cell lines have lower basal expression of NAG-1 than other gliomas and normal astrocytes. Most primary human gliomas have very low levels of NAG-1 expression. NAG-1 basal expression appeared to inversely correlate with tumor grade in glioma. Aberrant promoter hypermethylation is a common mechanism for silencing of tumor suppressor genes in cancer cells. In glioblastoma cell lines, NAG-1 expression was increased by the demethylating agent, 5-aza-2′-deoxycytidine. To investigate whether the NAG-1 gene was silenced by hypermethylation in glioblastoma, we examined DNA methylation status using genomic bisulfite sequencing. The NAG-1 promoter was densely methylated in several glioblastoma cell lines as well as in primary oligodendroglioma tumor samples, which have low basal expression of NAG-1. DNA methylation at two specific sites (−53 and +55 CpG sites) in the NAG-1 promoter was strongly associated with low NAG-1 expression. The methylation of the NAG-1 promoter at the −53 site blocks Egr-1 binding and thereby suppresses Nag-1 induction. Treatment of cells with low basal NAG-1 expression with NAG-1 inducer also did not increase NAG-1. Incubation with a demethylation chemical increased Nag-1 basal expression and subsequent incubation with a NAG-1 inducer increased NAG-1 expression. We concluded from these data that methylation of specific promoter sequences causes transcriptional silencing of the NAG-1 locus in glioma and may ultimately contribute to tumor progression.
Gliomas are the most prevalent central nervous system tumor of adults. Approximately 25% of total brain tumors are thought to arise from a glial progenitor cell.1 Glioblastoma (World Health Organization (WHO) Grade IV) has the poorest differentiation of glioma and is the most malignant human brain tumor. The median survival time is 12 months, and 5-year survival rate is <3% despite advances in antitumor therapy.2
Genetic alterations in glioma have been extensively studied, including TP53,3–5PTEN6 and p16 (CDKN2A)/CDK4/Rb alterations,7, 8 loss of heterozygosity of chromosome 19q3, 9 and EGFR amplification.3, 10 Mutations in the TP53 tumor suppressor gene occur in about one-half of all human cancers. TP53 mutations are the first detectable genetic alteration in two-thirds of precursor low-grade diffuse astrocytomas, anaplastic astrocytomas and glioblastomas.3–5 Transforming growth factor-β (TGF-β) also plays a key role in malignant glioma progression. Exogenous expression of TGF-β receptors has been proposed as a therapeutic avenue in glioma.11, 12 Epigenetic changes, such as DNA methylation and histone modification favoring transcriptional silencing, have also emerged as important contributors to tumorigenesis. DNA hypermethylation of promoter DNA at the p16, sFRP1, GATA4, GATA5, O6-MGMT, p73 and TMS1 loci has been reported in multiple cancer types.13 Hypermethylation of TIMP3, EMP3, CST6, BIK and TSPYL5 has been detected in glioblastoma multiforme.14, 15
The nonsteroidal anti-inflammatory drug-activated gene (NAG-1), a divergent member of the TGF-β superfamily, is involved in tumor progression and development.16NAG-1 was mapped to 19p12.1-13.1 and subsequently given various names, including GDF15, MIC-1, PTGF-β, PLAB, PDF and PL74. Previous investigations on the regulation of NAG-1 expression revealed complex mechanisms that can be modulated by multiple cellular stressors, such as acute injury, inflammation, cancer and drugs known to have antitumorigenic, proapoptotic or anti-inflammatory activities.17–20 Many studies showed that the overexpression of NAG-1 in cancer cells results in growth arrest and increases in apoptosis.21 NAG-1 transgenic mice expressing human NAG-1 ubiquitously appear to be less susceptible to genetically or chemically induced intestinal cancer.22 The NAG-1 transgenic mice have significantly suppressed polyp transformation in the small intestine after treatment with azoxymethane or after introduction of mutant APC.23 These findings suggest that NAG-1 may act as a tumor suppressor, but the detailed mechanisms of action of NAG-1 in tumorigenesis have not been elucidated to date.
NAG-1 protein has been detected in the epithelium in the central nervous system and is reported to function as both a neurotrophic and a neuroprotective factor.19, 24 NAG-1 expression is markedly increased in many solid cancers, including prostate, colon, breast, melanoma, pancreatic and thyroid cancer.25, 26 In contrast, our previous study demonstrated that NAG-1 expression in glioblastoma cell lines was significantly lower than in benign glioma cells and normal human astrocytes.27 Strelau et al. also reported that primary glioblastoma have less NAG-1 expression compared to other gliomas.28 This observed decrease in the expression of NAG-1 in glioblastomas suggests modulation of expression—potentially through epigenetic mechanisms such as DNA methylation. Methylated CpG islands have been identified in distinct subgroups of glioma, but analysis of methylated CpG islands was not informative for NAG-1.29 In our report, we asked whether the NAG-1 locus might be regulated by epigenetic mechanisms in glioblastoma. Our data implicate DNA methylation and accompanying transcriptional silencing as a mechanism coupling NAG-1 expression to tumor growth characteristics in glioma.
Material and Methods
Cell Culture and Reagents
The glioblastoma multiforme cell lines T98G, U118MG, U138MG, U373, U87MG and A172, the glioma cell line Hs683 and the low-grade astrocytoma cell line SW1088 were purchased from the American Type Culture Collection (Manassas, VA). The NHA cell line was obtained from Clonetics (San Diego, CA). All glioblastoma cells and Hs683 cells were maintained in Dulbecco's modified Eagle's medium, containing 4.5 g/l D-glucose and L-glutamine from Gibco (Grand Island, NY) and supplemented with 1 mM MEM sodium pyruvate solution (Gibco), 10 μg/ml gentamicin (Gibco) and 10% heat-inactivated fetal bovine serum. The SW1088 and NHA cells were maintained in AGM Bullet Kit from Lonza (Walkersville, MD). 5-Aza-2′-deoxycytidine (5-AZA-dC) and trichostatin A (TSA) were purchased from Sigma-Aldrich (St. Louis, MO) and were dissolved in 0.1% Me2SO (DMSO). Sulindac sulfide (SS) was purchased from Sigma and was dissolved in ethanol. Antibodies were from the following sources: anti-β-actin (1:2,000 dilution) was from Cell Signaling Technology (Beverly, MA). NAG-1 polyclonal rabbit antibody (1:1,500 dilution) was generated in this laboratory.21 Anti-Egr-1 (sc-110X) and normal rabbit IgG (sc-2027) were from Santa Cruz Biotechnology (Santa Cruz, CA).
Primary glioma tumor samples
All primary tumor samples used for this study were derived from snap-frozen surgical tissue samples from the Midwestern division of the cooperative human tissue network, USA.
Western blot analysis
Total cell lysates from cultured cells and primary tumor samples were isolated in RIPA buffer with Complete Mini protease inhibitor cocktail tablets from Roche (Indianapolis, IN), sonicated briefly and quantitated by BCA assay from Pierce (Rockfold, IL). Thirty micrograms of total protein per lane was separated by SDS-PAGE 4–12% Bis–Tris gel from Invitrogen (Carlsbad, CA) and transferred onto a nitrocellulose membrane (Invitrogen). The blots were blocked for 1 hr in 5% skim milk in Tris-buffered saline containing 0.05% Tween-20 (TBS-T; Sigma-Aldrich) and probed overnight at 4°C in 5% skim milk in TBS-T with each primary antibody. After washing with TBS-T, the blots were incubated with the appropriate horseradish peroxidase-conjugated secondary antibody for 1 hr at room temperature in 3% skim milk in TBS-T, washed several times in TBS-T and detected by the Amersham ECL plus Western blot detection system from GE Healthcare (UK).
To examine Nag-1 levels in the cell culture media, we used the human GDF-15 Duo Set ELISA Development kit from R&D systems (Minneapolis, MN) following the manufacturer's protocol. Briefly, an ELISA plate was coated with 2.0 μg/ml of monoclonal anti-human GDF-15 capture antibody. Then, various concentrations of standard recombinant GDF-15 and samples added to each well, and the plate was incubated for 2 hr at room temperature. After the plate was washed, 50 ng/ml of biotinylated goat GDF-15 antibody was incubated for 2 hr at room temperature. Finally, streptavidin-conjugated horseradish peroxidase was added, and the optical density of each well was determined using a microplate reader set to 450 nm.
Isolation and reverse transcription
Total RNAs from cultured cells and primary tumor samples were isolated with QIAshredder and RNeasy MINI kit from Qiagen (Valencia, CA) according to the manufacturer's protocol. One microgram of RNA was treated with 1 U of DNase I Amplification Grade (Invitrogen) at room temperature for 15 min to remove genomic DNA and followed by inactivation with 2.5 mM EDTA (Invitrogen) at 65°C for 5 min. Reverse transcription (RT) was performed with Super Script II Reverse Transcriptase (Invitrogen) according to the manufacturer's protocol. The cDNA was stored at −20°C until use.
Real-time PCR assays were performed using the iQ5 and MyiQ Real PCR detection system (Bio-Rad, Hercules, CA). Real-time RT-PCR fluorescence detection was performed in 96-well plates with iQ SYBR Green Supermix (Bio-Rad). Amplification primers were 5′-TGCCCGCCAGCTACAATC-3′ (forward) and 5′-TCTTTGGCTAACAAGTCATCATAGGT-3′ (reverse) for NAG-1 gene, and 5′-CCTGGCACCCAGCACAAT-3′ (forward) and 5′-GCGGATCCACACGGAGTACT-3′ (reverse) for β-actin. The threshold cycle number (CT) for same amount of cDNA of each sample was determined in triplicate. The copy number for NAG-1 was calculated from the CT value. All the results represent means ± SD of three independent experiments.
DNA extraction and bisulfite genomic sequencing PCR
Total genomic DNA from cultured cells and primary tumor samples were isolated with DNeasy Blood and Tissue Kit (Qiagen) according to the manufacturer's protocol. Bisulfite modification was done using the EZ DNA methylation Gold Kit (Zymo Research) following the manufacturer's instructions. Bisulfite sequencing PCR primers were designed with the Web software program EpiDesigner (http://www.epidesigner.com/). The sequences of the bisulfite sequencing PCR primers were 5′-AAGAGGATATTGAGGTTTAGAAATGTG-3′ (forward) and 5′-AATCTTCCCAACTCTAATTAACCC-3′ (reverse). This primer set was used to amplify the CpG island around the transcription start site of the NAG-1 gene (−217 to +258) in the promoter region with an expected 465-bp product. PCR reactions were done with Platinum Taq DNA polymerase (Invitrogen). The PCR products were resolved on a 2% agarose gel, purified with a QIAquick Gel Extraction Kit (Qiagen), cloned with TOPO XL PCR Cloning kits (Invitrogen) and sequenced with an ABI 3100 automated sequencer (Applied Biosystems).
Chromatin immunoprecipitation assay
Chromatin immunoprecipitation (Chip) assay was performed using the Chip assay kit from Upstate Biotechnology (Lake Placid, NY) according to the manufacturer's protocol. Briefly, T98G and U118MG cells (3 × 106) in the 10-cm dish were treated with 30 μM SS or control (0.1% Me2SO) for 4 hr and then fixed by the addition of 1% formaldehyde to media for 5 min at 37°C. The fixed cells were scraped into conical tubes, pelleted and lysed in SDS lysis buffer containing protease inhibitor (Roche and Sigma). DNA was sheared to fragments of 200–800 bp by sonication ten times for 30 sec. The sonicated cell supernatant was diluted tenfold in the Chip dilution buffer. The chromatin was precleared with salmon sperm DNA/protein A-agarose slurry for 1 hr at 4°C. The precleared supernatant was incubated with antibodies against Egr-1 (sc-110) or normal rabbit IgG overnight at 4°C. The immunocomplexes were eluted with elution buffer (1% SDS and 0.1 M NaHCO3). NaCl (5 M) was added into eluted samples to reverse histone–DNA crosslinks, and the samples were heated overnight at 65°C. Then, DNAs were purified and analyzed by quantitative PCR, using primers specific to sequences at the NAG-1 promoter, i.e., 5′-CAGCTGTGGTCATTGGAGTG-3′ and 5′-GAGCTGGGACTGACCAGATG-3′ for Sp-1-binding site BC.
The NAG-1 expression vector (pcDNA3-NAG-1) was described previously.21 U87MG cells were seeded in six-well plates at 30 × 104 cells per well in DMEM and grown to 60% confluence. One microgram of pcDNA3-NAG-1 or pcDNA3.1 (empty vector for expression control) was transfected using FuGENE 6 Transfection Reagent (Roche) according to the manufacturer's protocol. After 24-hr transfection, the efficiency of transfection was confirmed by Western blot and ELISA.
After transfection of pcDNA3.1 or pcDNA3-NAG-1, the cells were incubated with serum-free media for 5 days. Cells were trypsinized and harvested by centrifugation. Pellets were washed with PBS, resuspended in PBS containing 10 mg/ml propidium iodide (Roche) and 0.5 mg/ml RNase (Roche) and stored overnight at 4°C. Stained cells were analyzed using a Becton Dickinson FACSort (Franklin Lakes, NJ) as per manufacturer's instructions. The fraction of sub G1 cells in the population was determined using CellQuest 3.1 software (Becton Dickinson Flowcytometry System).
Colony formation in soft agar assay
After transfection of pcDNA3.1 or pcDNA3-NAG-1, the U87 cells were trypsinized and resuspended at 2 × 103 cells in 1 ml of 0.35% agar solution in growth media and then layered on top of a 0.7% agar layer in six-well plates. Plates were incubated for 2 weeks at 37°C in a 5% CO2 humidified atmosphere. Cell colonies were visualized following an overnight stain with 0.5 ml of 0.025% crystal violet (Sigma-Aldrich) and examined microscopically. These were represented as mean colony number examined in five randomly chosen microscope fields.
All statistical differences between experimental groups were evaluated by the two-tailed unpaired Student's t-test.
NAG-1 basal expression in glioma cells and primary tumors
The basal expression of Nag-1 in several glioma cell lines, normal human astrocytes and several primary glioma samples were determined at both the mRNA and protein levels. The basal expression of NAG-1 proprotein was very low in all glioblastoma cell lines except in A172. In contrast, low-grade glioma cell lines, such as Hs683 and SW1088, and normal human astrocytes exhibited higher levels of NAG-1 proprotein (Fig. 1a). NAG-1 proprotein is cleaved and secreted into the media as a mature dimer. Thus, the secreted form was measured in cell media to further analyze total NAG-1 formation. The basal expression of secreted NAG-1 protein in media was also very low in high-grade glioblastoma cell lines (U118, U138 and U373, Fig. 1b). A higher secreted NAG-1 was observed in T98G and A172 cells. In the low-grade cell lines and in the normal human astrocytes, high levels of the secreted protein were observed. Thus, in cultured cell lines, the high-grade glioblastomas express NAG-1 poorly.
We next measured NAG-1 protein levels in primary tumor samples. The basal expression of NAG-1 proprotein in all glioma samples was very low with a single exception (tumor 3, Fig. 1a). Further, NAG-1 expression was remarkably low in oligodendrogliomas (WHO Grade II) (tumor 10 to tumor 12, Fig. 1a). To confirm the low expression of NAG-1 in high-grade glioma, we measured the mRNA copy number. The expression was the lowest in the high-grade glioblastoma cell lines and glioma, whereas higher basal expression was observed in the low-grade or normal cell lines. This data suggest that the basal expression of NAG-1 is suppressed in higher grade glioma.
Methylation status of NAG-1 in glioma cell lines and primary tumors
The methylation status of CpG sites of the NAG-1 promoter (−162 to +55) (Table 1) and the intragenic CpG island (+97 to +234) (Supporting Information Table 1) was determined by bisulfite sequencing of genomic DNA isolated from the cell lines and primary tumors. In U118 and U138, the CpG sites of NAG-1 promoter were uniformly methylated. CpG dinucleotides at two sites (−53 and +55) of the NAG-1 promoter were more frequently methylated in the cells with low NAG-1 expression (U118, U138, U373 and U87) than in cells with moderate to high expression of NAG-1. The CpG site at position −53 of the NAG-1 promoter is Sp-1-binding site C, one of the pivotal sites regulating basal expression and is also critical to activated expression of the locus mediated by Egr-1.30 The CpG site at position +55 is very close to a putative p53-binding site that may also regulate NAG-1 expression.30 No correlation of DNA methylation status in the intragenic CpG island with expression was observed (Supporting Information Table). These results indicate that hypermethylation correlates with the low expression of NAG-1 in glioblastoma cell lines.
Table 1. Methylation status of NAG-1 promoter in glioma cells, NHA cells and primary tumor samples
Hypermethylation status was then measured in human glioblastoma samples. In primary glioblastoma (tumor 1 to tumor 8), these sites were less methylated than the corresponding site in glioblastoma cell lines, as expected. In anaplastic oligodendroglioma (WHO Grade III) (tumor 9) and oligodendroglioma (tumor 10 to tumor 12), CpG sites −118, −53, +17 and +55 of NAG-1 promoter were frequently methylated. The CpG site at position −118 of the NAG-1 promoter is Sp-1-binding site A, also a pivotal Sp-1-binding site. TSA induction of NAG-1 expression is dependent on Sp-1-binding site A and Sp-1-binding site C.27
In cell lines and primary tumor samples, the CpG island of the NAG-1 gene was evenly methylated. This methylation was limited to the +176 to +234 positions in normal human astrocytes. We concluded from these experiments that the suppression of NAG-1 expression is associated with DNA methylation of specific sites in the NAG-1 promoter.
Treatment with DNA demethylation agent induces NAG-1 expression
Epigenetic repression of tumor suppressor genes by DNA methylation and histone modification often disrupts differentiation of epithelial and hematopoietic cells and can promote cancer progression.31 Previously, we investigated the effect of the histone deacetylase inhibitor, TSA on NAG-1 regulation in glioma cells. TSA induced NAG-1 expression at the transcriptional and post-transcriptional levels in T98G glioblastoma cells. However, we did not observe the induction of NAG-1 by TSA in U87, U118, U138 and U373 glioblastoma cells, whose basal expression of NAG-1 was low27 and promoter hypermethylated. To investigate if the low response was due to hypermethylated promoter, U87, U118, U138 and U373 cells were treated with the DNA demethylation agent, 5-AZA-dC, and then basal NAG-1 expression was measured. The protein (Fig. 2a) expression of NAG-1 was significantly induced by 5-AZA-dC in three of four cell lines examined (U87, U118 and U373). Note that the exposure time on Figure 2a was longer than Figure 1a. We next measured the mRNA expression after incubation with 5AZA-dC or vehicle. Treatment with the demethylation drug increased the expression of NAG-1 mRNA in all cells tested except U138 with expression too low to accurately measure (Fig. 2b).
Sulindac sulfide treatment in different methylation status cell lines
The cyclooxygenase inhibitor is a potent inducer of NAG-1 mediated by the induction of Egr-1 that binds to Sp/Egr1 site in the NAG promoter. Methylation of this region may block the induction of NAG-1 by sulindac sulfide (SS). To test this hypothesis, T98G (lower methylation status) and U118 cells that have high methylation level were incubated with SS. The T98G cells that have a hypomethylated promoter showed good induction of NAG-1 protein by SS (Fig. 3a). In contrast, U118 cells that have a hypermethylated promoter showed poor induction of NAG-1 protein. We also measured secreted NAG-1 protein in medium at different time points after treatment (Fig. 3b). The T98G cells also showed significant induction of NAG-1, but the U118 cells showed very low expression.
Effects of combination treatment of 5-AZA-dC with SS or TSA
Following pretreatment for 5 days with 5-AZA-dC, T98G and U118 cells were passaged, starved overnight with serum-free medium and subsequently treated with vehicle, SS or TSA in serum-free medium for 24 hr, and the mature Nag-1 dimer was measured in the media. In T98G cells, pretreatment with 5-AZA-dC did not alter the respond to either NSAIDS (SS) or TSA. In contrast, in U118 cells, pretreatment with 5-AZA-dC significantly increased basal Nag-1 protein expression and increased the respond to both drugs (Fig. 3c). These results suggest that methylation of the NAG-1 promoter in glioblastoma cell lines prevents the NSAID-induced increases in Nag-1 protein observed in cells lacking promoter methylation.
Hypermethylation blocks Egr-1 binding to Nag-1 promoter
SS mediates increases in NAG-1 expression by an increase in Egr-1 expression. Egr-1 then binds to the Sp1/Egr-1 sites in the NAG-1 promoter, stimulating gene expression. SS induced Egr-1 mRNA expression in T98G and U118 cells (Fig. 4a) with the peak mRNA expression observed at 2 hr after treatment in both cell lines. We suspected that hypermethylation in the Sp-1/Egr-1 site would block Egr-1 binding to Nag-1 promoter. To test this hypothesis, chromatin immunoprecipitation assay was used to determine whether Egr-1 protein is bound to the NAG-1 promoter. T98G and U118 cells were treated with SS or vehicle for 4 hr. Subsequently, the Egr-1 protein and DNA complex were measured. SS treatment induced an approximately twofold increase in bound Egr-1 in T98G cells, but there was no increase in U118 cells (Fig. 4b). To confirm whether demethylation increases Egr-1 binding to NAG-1 promoter, following pretreatment for 5 days with 5-AZA-dC, T98G and U118 cells were passaged, starved overnight with serum-free medium and subsequently treated with SS or vehicle in serum-free medium for 4 hr, and then Egr-1 and DNA complex were measured. T98G showed about 2.5-fold increase and U118 showed about twofold increase by SS (Fig. 4c). Following treatment for 5 days with 5-AZA-dC, CpG sites −67, −53 and +55 of Nag-1 promoter were surely demethylated in U118 cell, which was confirmed by bisulfate sequencing of genomic DNA (Supporting Information Table 2). Taken together, these findings indicate that hypermethylation blocks Egr-1 binding to Nag-1 promoter and, therefore, suppresses the Nag-1 induction.
NAG-1 expression and apoptosis in glioma cells
NAG-1 expression is associated with cell cycle arrest and increased apoptosis in a number of cells, but this has not been studied in glioblastomas. Furthermore, we wondered if the expression of NAG-1 in glioma having a hypermethylated promoter would response to NAG-1 expression. The effect of NAG-1 overexpression was examined in several glioblastoma cell lines that have highly methylated promoters and hence low basal expression of NAG-1 (U87, U118, U138 and U373 cells). Each cell line was transfected with NAG-1 expression plasmid, and the NAG-1 expression was confirmed by Western blot for the proform (Fig. 5a) and ELISA assay for the secreted NAG-1 (Fig. 5b). NAG-1 proteins in both cell lysates and growth media in all cells were highly expressed in the transfected cells. Subsequently, cells were incubated with serum-free medium for 5 days, and then cell cycle was measured by flow cytometry. NAG-1 overexpression significantly increased the proportion of apoptotic cells as measured by the sub-G1 population in U87, U118 and U373 (Fig. 5c). Thus, NAG-1 expression has a proapoptotic effect in glioblastoma cells that have low basal expression of NAG-1. Because NAG-1 acts as a tumor suppressor, we next examined whether NAG-1 expression would inhibit the growth of glioblastomas on soft agar. U87 cells transfected with vector control or NAG-1 expression construct were grown in soft agar and colony formation measured. The expression of NAG-1 inhibited cell growth by ∼50% (Fig. 5d). These findings support the hypothesis that NAG-1 appears to act as a tumor suppressor gene in glioblastomas but is epigenetically silenced in high-grade tumors.
Gliomas are the most common central nervous system tumor in both adults and children. Although in vivo and in vitro studies have shed new light on the mechanism of tumorigenesis of neuroepithelial tumors, current treatments do not result in an improvement in the very poor prognosis of this disease. However, epigenetic studies have indicated possible strategies for tumor-specific and customized therapy.32–35 The restoration of silenced tumor suppressor genes is an attractive therapeutic approach that could result in improvements in prognosis.
In a previous study,27 we have investigated epigenetic modifications, including histone deacetylation and DNA methylation, ultimately identifying the NAG-1 gene as being epigenetically regulated and possibly silenced in gliomas. The histone deacetylase inhibitor, TSA, induces NAG-1 expression in several glioma cell lines including T98G cells. TSA induces apoptosis in T98G cells, and the increase in expression of NAG-1 plays an important role in TSA-mediated apoptosis, suggesting histone acetylation regulates the expression of NAG-1. However, 5-AZA-dC did not induce NAG-1 expression in T98G cells, suggesting methylation was not important in regulation of NAG-1 expression in these cells. In our report, we examined in more detail the methylation status of the NAG-1 promoter in glioblastoma cells and in human glioma tumors. The promoter region of NAG-1 was poorly methylated in cells that have high basal expression of NAG-1 and hypermethylated in cells that have low basal expression of NAG-1. In glioblastoma cell lines, an inverse relationship between the basal expression and promoter hypermethylation is clearly apparent with hypermethylation observed in the high-grade cells. The study of methylation status and basal NAG-1 expression was continued by examination of the basal NAG-1 expression and promoter methylation in human glioma samples. In general, we observed a very low basal NAG-1 expression in the tumors and a high level of promoter methylation. 5-AZA-dC induced NAG-1 expression in glioblastoma cell lines with low basal expression of NAG-1 and a highly methylated promoter but was not effective in cells like T98G that have high basal expression and a poorly methylated NAG-1 promoter. Thus, we conclude that NAG-1 is epigenetically regulated and silenced in glioblastomas by methylation.
In the human tumor samples, the relationship between hypermethylation and basal NAG-1 expression is not as clear. In cell lines where the NAG-1 promoter is extensively methylated, clusters of DNA methylation were identified that correlate with known transcription factor-binding sites responsible for the regulation of basal expression. NAG-1 expression is downregulated in greater than 90% of the tumors. However, hypermethylation at the sites identified in cell lines (the −118 to −53 regions) was not observed in all human glioblastoma samples. Our observation is similar to results reported for methylation of other genes in glioblastoma.36 As previously suggested,36 the decreased levels of methylation observed in tumors versus cell lines may reflect differences in cellularity (tumors have multiple cell types), differences in the signaling environment in vivoversus in culture or that DNA methylation in cell lines may reflect a growth advantage in culture not evident in tumors.
NAG-1 expression is regulated by several transcriptional factors and post-transcriptional mechanisms, indicating a diverse regulation by antitumorigenic compounds. From our studies on the regulation of NAG-1 expression by tumor prevention drugs, it was revealed that NAG-1 is an important downstream target of three tumor suppressor pathways, p53,30 Egr-137 and AKT/GSK-3β.38 These pathways involve specific sites located in the −133 to +55 region of NAG-1 promoter, which we analyzed for the methylation status. The Egr-1/Sp-1-binding site (−67 and −53, respectively) regions are pivotal regulation sites for increasing NAG-1 expression by COX inhibitors, troglitazone and TSA. In addition, these sites are critical in the regulation of the basal expression of NAG-1.23 Our study indicated that Egr-1/Sp-1-binding site was highly methylated in glioblastoma cell lines and primary oligodendroglioma samples that have low basal expression of NAG-1. In addition, our methylation status assay showed that +55 region located close to p53-binding site was also highly methylated. The p53 sites play an important role in the regulation of NAG-1 expression and are pivotal in induction of NAG-1 expression by dietary compounds, for example, DADS and resveratrol.30, 39 Thus, the −128 to −53 region and +55 region of NAG-1 promoter are likely involved in silencing of the tumor suppressor gene in tumorigenesis of glioma.
Methylation of the Egr-1/SP-1 site not only decreases the basal expression of NAG-1 but also prevents an increase in expression after treatment with SS or TSA. SS increased NAG-1 expression in T98G cells with a poorly methylated NAG-1 promoter, whereas treatment of U-118 cells that have a highly methylated promoter did not increase NAG-1 expression. After removal of the methyl group by 5-AZA-dC treatment, SS increased NAG-1 expression in the U-118 cells. Although SS increased in both the T98G and U-118 cells the expression of Egr-1, a transcription factor required for the increase in transcriptional activity by SS, the binding of Egr-1 to the promoter site was blocked on the methylated promoter in U-118 cells as determined by the Chip assay. Thus, in glioma, hypermethylation of NAG-1 promoter silenced basal expression and blocked drug-induced expression, a finding that may provide insight into why gliomas are resistant to many drug therapies.
The role for NAG-1 in development of tumors is highly complex. Studies in a number of tumor cells, colorectal,21 breast18 and now glioblastoma all indicate that the expression of NAG-1 induced apoptosis, inhibited the growth on soft agar and altered the cell cycle. Other studies in nude mice show that the expression of NAG-1 inhibits the growth of tumors in xenographs.21 NAG-1 transgenic mice expressing the human NAG-1 are resistant to both chemical and genetically induced intestinal tumors.22 Our current studies support the hypothesis that NAG-1 is a potential tumor suppressor gene and results of this investigation confirm that its expression can be regulated by gene hypermethylation in glioblastoma. Further studies are needed to determine whether NAG-1 transgenic mice are resistant to development of glioblastoma and if hypermethylation silences NAG-1 expression in other cancers.