Over the last decade, there has been major progress in the identification and characterization of human tumor antigens recognized by the host immune system. A subgroup of tumor antigens, commonly referred to as cancer/testis antigens (CTAs), are expressed only in the tissues of the testis, ovary and placenta under normal conditions, but are also expressed in various types of human tumors.1, 2 Since normal CTA-expressing tissues do not express major histocompatibility complex (MHC) class I molecules, CD8 T cells cannot recognize CTAs expressed on these tissues, suggesting that CTAs are the ideal targets for tumor immunotherapy. CTAs and genes were originally identified through a variety of methods. These include T-cell epitope labeling,3 serological analysis of cDNA expression libraries (SEREX),4, 5 differential gene expression analysis6 and bioinformatics methods.7–9 In particular, NY-ESO-1 is the most immunogenic CTA discovered thus far, and it is considered to be a highly promising therapeutic target for immunotherapy.10 To date, very little is known regarding the physiological function(s) of these antigens or the mode in which the expression of their gene families is regulated.
Epigenetic alterations, including hypermethylation of promoter CpG islands, histone deacetylation of tumor suppressor and tumor-related genes,11–13 and global DNA hypomethylation,14, 15 have been recognized as important contributors to carcinogenesis in humans. Global DNA hypomethylation has been observed in various neoplasms and is considered to occur at the early stages of tumor development.16–18 However, it has been shown that the expression of certain CTA genes is reactivated in cancerous cells; this could be due to a loss of epigenetic regulation as observed when methylated chromatin regions are demethylated or when deacetylated histones are acetylated.19 Therefore, recent evidence shows that the deregulation of the DNA methylation apparatus that occurs during cancer development could provide new therapeutic targets for cancer treatment.
The DNA demethylating agent 5-aza-2′-deoxycytidine (5-aza-CdR, decitabine) is a cytosine analogue that is incorporated into DNA during replication. It covalently binds DNA methyltransferase and inhibits its activity, leading to genome-wide demethylation.20–24 There have been several studies demonstrating the ability of 5-aza-CdR to activate the gene expression of CTAs in vitro and in vivo, which may be silenced by the hypermethylation of their promoters. This drug has been used in clinical studies for the treatment of chronic myelogenous leukaemia (CML), sickle cell anaemia and myelodysplastic syndrome (MDS).20, 25–27 Previous evidence has clearly defined the epigenetic regulatory role of DNA methylation in the constitutive expression of CTAs by cutaneous melanoma cells and renal cancer cells and has demonstrated that in vitro treatment with 5-aza-CdR upregulated their expression in neoplastic cells.28, 29
Gliomas are the most common primary tumors of the central nervous system; they account for 30% of adult primary brain tumors. Brain tumors remain difficult to cure despite recent advances in surgical, radiotherapeutic and chemotherapeutic approaches. In particular, there is currently no optimal treatment for glioblastoma multiforme, the most common malignant brain tumor in adults, and patients typically survive for a period less than a year. The poor outcome partly relates to the inability in delivering chemotherapeutic agents through the blood–brain barrier (BBB) and the low effect of radiation on the tumor. Therefore, new and more effective strategies are urgently required. Of these, the establishment of immunotherapy specifically targeting malignant cells is expected to improve tumor prognosis. It has recently been demonstrated that malignant glioma cells express certain known tumor-associated antigens such as HER-2, gp100, MAGE-1 and IL-13 receptor α2.30–37 However, CTA profiling of glioma cells, in particular, in the Asian population remains unknown.
In our study, we analyzed the expression of 13 CTA genes (MAGE-1, MAGE-3, MAGE-4, MAGE-6, MAGE-10, MAGE-3/6, LAGE-1, CT7, SCP-1, SSX-1, SSX-2, SSX-4 and NY-ESO-1) in 30 glioma tissues, 5 human glioma cell lines and 3 newly established cell lines from the Japanese population. Subsequently, the role of 5-aza-CdR in the regulation of the expression of various CTAs in glioma cells was analyzed. Finally, we demonstrated that NY-ESO-1, one of the most antigenic CTAs, is effectively induced in human glioma cells by 5-aza-CdR, and that these glioma cells forcibly expressing NY-ESO-1 show in vitro and in vivo sensitivity to NY-ESO-1-specific cytotoxic T lymphocytes (CTLs). Our present study provides the basis to establish novel immunotherapeutic approaches in glioma patients.
Material and methods
The human glioma cell lines U251 (human leukocyte antigen (HLA)-A2), SKMG1 (HLA-A24), AO2 (HLA-A3), U87MG (HLA-A2) and T98 (HLA-A2) were obtained from the Memorial Sloan-Kettering Cancer Institute (New York, NY) and maintained in Eagle's minimal essential medium (MEM) at 37°C in a humidified atmosphere of 5% CO2 in air. A human osteosarcoma cell line, namely, SaOS-2 (NY-ESO-1+, HLA-A2), was kindly provided by Dr. Y. Nishida (Department of Orthopaedics, Nagoya University, Nagoya, Japan) and maintained in Dulbecco's modified Eagle's medium (DMEM; Nissui, Tokyo, Japan). A human myeloid leukaemia cell line K562 and the HLA-A*0201-transfected T2 cell line (T2.A2) were maintained in RPMI 1640 (Invitrogen, Carlsbad, CA). The medium was supplemented with 10% fetal bovine serum (FBS), 5 mM L-glutamine, 2 mM nonessential amino acids, and antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin). Commercially available normal human astrocytes (NHA; Cambrex, Baltimore, MD) maintained in AGM medium with BulletKit supplement (Cambrex), human aortic smooth muscle cells (AoSMC; Cambrex), human adult fibroblasts (NHDF-ad; Cambrex) maintained in DMEM with 10% FBS, and human epidermal keratinocytes (NHEK, Kurabo, Osaka, Japan) maintained in Epilife medium (Cascade Biologics, Nottinghamshire, UK) supplemented with HuMedia-KG (Kurabo) were used as normal cells.
Reagents and peptides
A vial containing 5 mg lyophilized powder of 5-aza-CdR was obtained (Sigma-Aldrich, St. Louis, MO). The vial was reconstituted with 20 ml sterile water to obtain a 1 mM solution, and it was stocked at 4°C. The HLA-A2-binding NY-ESO-1 peptide, p157-165 (SLLMWITQC), which was initially identified using the T-cell line NW38-IVS-138 and the HLA-A2-binding IL-13Rα2 peptide, p345-354 (WLPFGFILI), which was identified previously,37 was synthesized by Thermo Electron GmbH (>90% purity) (Ulm, Germany), and solubilized in 50% dimethyl sulphoxide (DMSO)/water.
Collection of surgical specimens
Thirty tumor samples were collected at the Nagoya University School of Medicine from patients whose tumors were histologically diagnosed as gliomas (WHO Grade II, III and IV). Genetic analysis in our study was approved by the ethical committee of Nagoya University Hospital. All patients provided written informed consent. All tissues were frozen immediately and stored at −80°C until use.
Primary-culture cells derived from patients with high-grade gliomas
Tumor tissues were derived from 3 patients with high-grade gliomas who had undergone surgical resection in Nagoya University Hospital, Nagoya, Japan; the tissues were primary-cultured as follows. Immediately after the brain tumors were removed from the patients, the tissues were homogenized and digested with 1% DNase and 0.1% trypsin for 30 min at 37°C and centrifuged at 800 rpm for 5 min. The cells were seeded at a density of 2 × 106 cells per 100-mm dish and maintained in DMEM supplemented with 10% FBS, 5 mM L-glutamine, 2 mM nonessential amino acids, and antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin). The cells were then incubated in a standard tissue culture incubator (100% relative humidity and 5% CO2). After achieving 80–90% confluence, they were subcultured onto a new 100-mm plate at a density of 2 × 106. The established cell lines were designated as NNS-10, NNS-11 and NNS-12. All cells were immunocytochemically confirmed as glioma cells based on their expression of glial-fibrilliary acidic protein (GFAP).
In vitro treatment of cultured cells with 5-aza-CdR
Treatment with 5-aza-CdR was performed as described in a previous study.29 Cells were seeded at a density of 1.0 × 105 cells/well in a 6-well plate, and placed at 37°C overnight in a 5% CO2 incubator. The next day, the cell culture medium was replaced with fresh medium containing 0.1, 1 and 10 μM 5-aza-CdR. The medium was changed every 12 hr for 2 days. The plates were wrapped in aluminum foil to avoid light exposure. At the end of the treatment, the medium was replaced with fresh medium without 5-aza-CdR, and the cells were cultured for an additional 48 hr.
RNA was isolated from 25 to 50 mg of tissue/sample by using Trizol (Invitrogen) according to the manufacturer's protocol. RNA was finally resuspended in nuclease-free water.
Conventional reverse transcription PCR
cDNA was synthesized from 1 μg total RNA by using random hexamers and the Superscript II reverse transcription (RT) kit (Invitrogen), according to the manufacturer's protocol. For each PCR, 2 μl cDNA was used in a 20-μl reaction volume containing 200 mM dNTPs, 1 mM sense and antisense primers, 1.25 mM MgCl2, 2 μl 10× PCR buffer, and 0.5 U Taq polymerase (Applied Biosystems, Foster City, CA). The PCR primers used were those listed in the Ref.39 β-actin was used as the endogenous control. The cycling parameters were as follows: denaturation at 95°C for 45 s, primer annealing at 55°C for 45 s, and 35 cycles of extension at 72°C for 60 s. PCR cycling was preceded by an initial denaturation at 95°C for 5 min, followed by final extension at 72°C for 5 min. The PCR products were analyzed by electrophoresis on a 1.5% agarose gel; this was followed by staining with ethidium bromide. If no signal was observed, the remaining PCR products were amplified by an additional 15 cycles and analyzed again to confirm the absence of the signal.
Quantitative RT-PCR for NY-ESO-1 expression
After synthesis of the first cDNA strand as described earlier, quantitative RT-PCR was performed on the LightCycler real-time RT-PCR system (version 3.39) (Roche, Mannheim, Germany), using the FastStart Taqman Probe Master (ROX) (Roche) along with sets of primers and Universal ProbeLibrary probes (Roche) designed online with ProbeFinder version 2.40 (Roche). Probes specific for NY-ESO-1 were as follows: forward primer: 5′-TGTCCGGCAACATACTGACT-3′, reverse primer: 5′-ACT- GCGTGATCCACATCAAC-3′, and Universal ProbeLibrary probe Human No. 67 (Roche), which yields a 111-nt amplicon. The endogenous reference gene GAPDH was amplified by the forward primer: 5′-AGCCACATCGCTCAGACAC-3′, reverse primer: 5′-CGCCCAATACGACCAAATC-3′, and Universal ProbeLibrary probe Human No. 60, which yields a 67-nt amplicon. Each sample was amplified as follows: 1 cycle at 95°C for 10 min followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. cDNA from SaOS-2 was used to generate the standard curves for NY-ESO-1 and GAPDH. Standardization of samples was achieved by dividing the copy number of the target gene NY-ESO-1 by that of the GAPDH gene. Values were expressed as ratios relative to NY-ESO-1 expression in SaOS-2.
Western blot analysis
Cells treated with 5-aza-CdR (1 μM) and untreated cells were lysed in a buffer containing protease inhibitors. Protein samples (45 μg) were denatured at 100°C for 5 min and subsequently applied to each well and electrophoresed on a 12.5% polyacrylamide gel. After transferring the proteins to a polyvinylidene difluoride (PVDF) membrane, it was blocked with 3% low-fat skim milk, incubated with a monoclonal antibody (mAb) against NY-ESO-1 (ES121; Zymed, San Francisco, CA), and then washed and incubated with an horseradish peroxidase (HRP)-labelled secondary Ab. Visualization was performed using an enhanced chemiluminescence technique.
NY-ESO-1 promoter DNA methylation analyses
Methylation-specific PCR (MSP)40 and Pyrosequencing™ technology41 were used to determine the methylation status of the CpG island region of NY-ESO-1. None of the sequences identified in the NY-ESO-1 promoter region fulfilled the criteria for CpG islands (length, >200 bp, G + C content, >50%, observed CpG/expected CpG ratio, ≥0.6).42 The CpG islands and the location of the primers in exon 1 for DNA methylation analyses are shown in Figure 1. Genomic DNA isolation and sodium bisulphite conversion were performed as described previously.43 The primer sequences and annealing temperatures of MSP were as reported previously.44 The primer sequences for Pyrosequencing™ were designed using PSQ Assay Design (Biotage, Uppsala, Sweden). A 50-μl PCR was carried out in 60 mM Tris ± HCl (pH 8.5), 15 mM ammonium sulphate, 2 mM MgCl2, 10% DMSO, 1 mM dNTP mix, 1 U Taq polymerase, 5 pmol forward primer (5′-GGGTTGAATGGATGTTGTAG-3′), 50 pmol reverse primer (5′-CRCCACCAAAACTATCAA-3′), 50 pmol biotinylated universal primer (5′- GGGACACCGCTGATCGTTTACRCCACCAAAACTATCAA-3′) and ∼50 ng bisulphite-treated genomic DNA. The forward primer contains a 20-bp linker sequence on the 5′ end that is recognized by biotin-labeled primera; thus, the final PCR product can be purified using Sepharose beads. PCR cycling conditions were as follows: 95°C for 30 s, 60°C for 45 s and 72°C for 45 s for 55 cycles. The biotinylated PCR product was purified and made single-stranded to act as a template in a pyrosequencing reaction using the Pyrosequencing Vacuum Prep Tool (Pyrosequencing, Westborough, MA), as recommended by the manufacturer. In brief, the PCR product was bound to Streptavidin Sepharose HP (Amersham Biosciences, Uppsala, Sweden), and the Sepharose beads containing the immobilized PCR product were purified, washed, denatured using 0.2 M NaOH solution and washed again. Then, 0.3 mM pyrosequencing primer (5′-TGAATGGATGTTGTAGATG-3′) was annealed to the purified single-stranded PCR product, and pyrosequencing was performed using the PSQ HS 96Pyrosequencing System (Pyrosequencing). Methylation quantification was performed using the provided software.
Flow cytometric analysis
Glioma cells were treated with 5-aza-CdR as described earlier. The treated and untreated cells were stained with anti-HLA class-I mAbs (W6/32; kindly provided by Dr. K. Itoh, Kurume University, Kurume, Japan) or isotype control mouse IgG2a followed by staining with fluoroscein isothiocyanate (FITC)-labelled anti-mouse IgG Abs. Flow cytometric analysis of stained cells was performed by using FACS Calibur (Becton Dickinson, San Jose, CA).
Induction of HLA-A2-restricted NY-ESO-1-specific CTL lines by peptide-pulsed dendritic cells
Peripheral blood mononuclear cells (PBMCs) from healthy volunteers genetically typed as HLA-A2-positive by SRL, (Tokyo, Japan) were isolated using Ficoll-Paque (Amersham Biosciences). Peptide-pulsed dendritic cells (DCs) were prepared from donor-derived PBMCs, as described previously37 with minor modifications. In brief, the plastic adherent cells from PBMCs were cultured in AIM-V medium (Life Technologies, Gaithersburg, MD) supplemented with recombinant human granulocyte/macrophage-colony stimulating factor (rhGM-CSF, 500 U/ml) and rhIL-4 (500 U/ml) at 37°C in a humidified CO2 (5%) incubator. After 6 days, the culture medium was removed, and the immature DCs were cultured in AIM-V supplemented with rhGM-CSF (500 U/ml), recombinant human interleukin (rhIL)-4 (500 U/ml), rhIL-6 (1,000 U/ml), recombinant human tumor necrosis factor-α (10 ng/ml), and IL-1β (10 ng/ml). All cytokines used were purchased from Strathmann Biotech AG, Hanover, Germany. Mature DCs were harvested after another 2 days, resuspended in AIM-V medium at 1 × 106 cells/ml with the peptide (10 μg/ml), and incubated for 4 hr at 37°C. The peptide-pulsed DCs were then treated with mitomycin-C, washed, and finally resuspended in AIM-V medium supplemented with 10% human AB serum. Autologous CD8+ T cells were enriched from PBMCs by using magnetic microbeads (Miltenyi Biotech, Auburn, CA) and were added (1 × 106/well) to the peptide-pulsed DCs (1 × 105/well) in 2 ml AIM-V medium supplemented with 10% human AB serum, 1,000 U/ml rhIL-6, and 10 ng/ml rhIL-12 in each well of 24-well tissue culture plates. On day 7, the lymphocytes were restimulated with mitomycin-C-treated autologous DCs pulsed with peptides in AIM-V medium supplemented with 10% human AB serum, rhIL-2 and rhIL-7 (10 U/ml each). One week later and on a weekly basis thereafter, the responder cells were restimulated with peptide-pulsed DCs in a medium supplemented with rhIL-2 and rhIL-7 (10 U/ml each). The induction of NY-ESO-1-specific CTL lines was attempted 3 times.
The susceptibility of the untreated and 5-aza-CdR-treated U251 glioma cells (HLA-A2) to HLA-A2-restricted NY-ESO-1-specific CTL lines was evaluated by a standard 4-h 51Cr-releasing assay at various (effector:target) E:T ratios. The percentage specific lysis was calculated as follows: 100 × (experimental release − spontaneous release)/(maximum release − spontaneous release).
Cold target inhibition assay
The cold target inhibition assay was performed as described previously.45 In brief, T2.A2 cells were incubated with the NY-ESO-1 peptide or the irrelevant IL-13Rα2 peptide at a concentration of 10 μM for 1 hr. After extensive washing, the indicated numbers of peptide-loaded cells were incubated with 2 × 104 cytotoxic effector cells for 1 hr, and then 2 × 10351Cr-labeled, 5-aza-CdR-treated U251 glioma cells were added to each well. Cytotoxicity was assessed at the E:T ratio of 10:1 as described earlier.
Orthotopic glioma xenograft model
Female nonobese diabetic/severe combined immunodeficiency disease (NOD/SCID) mice aged 5 weeks were used in the experiments. All mice were purchased from SLC, Hamamatsu, Japan. They were maintained under specific pathogen-free conditions in the animal facility of Nagoya University School of Medicine. Animal experiments were performed according to the principles enunciated in the Guide for the Care and Use of Laboratory Animals prepared by the Office of the Prime Minister of Japan.
Mice were anaesthetized with an intraperitoneal injection of pentobarbital (60–70 mg/kg body weight). After shaving the hair and incising the scalp, a burr hole was made in the skull 3 mm lateral to the midline and 4 mm posterior to the bregma using a dental drill. The head of each mouse was fixed in a stereotactic apparatus with ear bars, and 2 × 105 U251 cells in 2 μl PBS were stereotactically injected over 4 min at a depth of 2 mm below the dura mater. A sterile Hamilton syringe fitted with a 26-gauge needle was used with a microsyringe pump. The needle was retained in the brain for an additional 2-min duration and then slowly withdrawn.
To test the induction of NY-ESO-1 in orthotopic glioma xenografts, 5-aza-CdR [0.2 mg/kg, intraperitoneally (i.p)] was administered every 12 hr for 4 consecutive days (8 pulses) starting on day 14 after glioma inoculation. The control brain tumor model was treated under similar experimental conditions by an i.p. PBS injection. To determine whether the treated cells upregulated the NY-ESO-1 expression, the mice were killed on day 3 after the final 5-aza-CdR injection. The brain tumors were removed for RT-PCR for NY-ESO-1.
Tumor-inoculated NOD/SCID mice were randomly divided into 4 treatment groups (10 animals each) to examine the treatment efficacy of NY-ESO-1-specific CTLs in combination with 5-aza-CdR. The animals in the various groups were then injected as follows: Group I, PBS (i.p., 8 pulses starting from the day after tumor inoculation) and PBS (2 μl) intratumorally (i.t.) 6 days after tumor inoculation; Group II, PBS i.p. (8 pulses) and bulk CTLs (i.t., 1 × 106 cells/mouse in 2 μl); Group III, 5-aza-CdR i.p. (0.2 mg/kg, 8 pulses) and PBS (i.t., 2 μl); Group IV, 5-aza-CdR i.p. (0.2 mg/kg, 8 pulses) and bulk CTLs (i.t., 1 × 106 cells/mouse in 2 μl). Five mice in each group were euthanized using CO2 inhalation on day 21 after tumor injection. The mice were transcardially perfused with 4% paraformaldehyde. The brain tissue was postfixed overnight, embedded in paraffin, and then cut into 5-μm serial horizontal sections. Tissue sections were stained by the standard haematoxylin and eosin technique. The growth inhibitory effect was evaluated by measuring the long (a) and short (b) axes in the coronal section showing the maximal area of each tumor. The approximate volume of the tumor (V) was calculated according to the formula, V (mm3) = a × b2/2. In addition, the survival times of all the remaining mice were recorded.
The statistical significance of the difference in tumor volumes and Kaplan–Meier survival curves was determined by analysis of variance (ANOVA) (StatView, SAS Institute, Cary, NC) with Bonferroni's correction for multiple comparisons and the log-rank test (StatView), respectively.
Expression of individual CTA genes in human gliomas and glioma cell lines
Although the expression frequencies of many CTAs in a variety of neoplasms have been determined, their expression in human gliomas remains unclear, and the expression frequency of CTAs varies drastically among ethnic groups.39 It would be useful to analyze the expression pattern of CTAs in gliomas in the Japanese population. In our study, RT followed by 35 cycles of amplification revealed that the expression of CTA genes was nearly imperceptible in human gliomas, and the expression pattern of these genes did not correlate with the histological grades of the gliomas (Table I). MAGE-4, MAGE-10 and NY-ESO-1 were expressed in only 1 of 30 samples, and MAGE-1 and MAGE-3/6 were expressed in 2 samples. The samples were negative for MAGE-3, MAGE-6, LAGE-1, CT7, SCP-1, SSX-1, SSX-2 and SSX-4 even after 50 amplification cycles of the 30 brain tumor specimens tested (Fig. 2a and Table I). Moreover, the human glioma cell lines and primary-cultured glioma cells did not test positive for any of the 13 CTAs even after 50 amplification cycles (Table II). The only exception was LAGE-1; SKMG1 and T98 cells constitutively expressed LAGE-1.
Table I. RT-PCR Analysis for Expression of Cancer-Testis Antigens in Glioma Tissues
We investigated the composite expression of 13 CTAs in 30 human glioma specimens by RT-PCR analysis. MAGE-4, MAGE-10 and NY-ESO-1 were expressed in only 1 sample out of 30, and MAGE-1 and MAGE3/6 were expressed in 2 samples. Other CTAs were all negative in the 30 brain tumor specimens tested. The expression pattern of these did not correlate with the histological grades of gliomas.
Black and white boxes indicate positive and negative expression, respectively.
Table II. Induction of CTA Expression by 5-aza-CdR in Glioma Cells and Normal Cells
We assessed whether CTAs could be induced by 5-aza-CdR treatment in 5 human glioma cell lines, 3 primary glioma cell lines and 4 human normal cells. The cells were treated with 0.1, 1 and 10 μM 5-aza-CdR every 12 h for 2 consecutive days (4 pulses). The human glioma cell lines and primary-cultured glioma cells did not test positive for any of the 13 CTAs even after 50 amplification cycles. The only exception was LAGE-1; SKMG1 and T98 cells constitutively expressed LAGE-1. Exposure to 5-aza-CdR invariably induced the expression of LAGE-1, MAGE-1, MAGE-3, MAGE-4, MAGE-3/6, SCP-1, SSX-1, SSX-2, SSX-4 and NY-ESO-1 in CTA-negative glioma cells. Because of toxicity, NY-ESO-1 expression in SK-MG-1, and MAGE-3/6 and LAGE-1 expression in U87MG are not observed when treated with 10 μM 5-aza-CdR. In contrast to glioma cells, administration of the agent does not induce CTA expression in human astrocytes, fibroblasts, smooth muscle cells and epidermal keratinocytes.
Effect of 5-aza-CdR on glioma cells and normal human cells in vitro
We assessed whether CTAs could be induced by 5-aza-CdR treatment in 5 human glioma cell lines and 3 primary glioma cell lines. The glioma cells were treated with 0.1, 1 and 10 μM 5-aza-CdR every 12 hr for 2 consecutive days (4 pulses). Table II summarizes the expression of the 13 CTA genes detected by RT-PCR in glioma cell lines treated with 0.1, 1 and 10-μM 5-aza-CdR. Exposure to 5-aza-CdR invariably induced the expression of LAGE-1, MAGE-1, MAGE-3, MAGE-4, MAGE-3/6, SCP-1, SSX-1, SSX-2, SSX-4 and NY-ESO-1 in CTA-negative glioma cells. Among these, we focused on NY-ESO-1-the most immunogenic CTA. Real-time quantitative RT-PCR was performed to quantitate NY-ESO-1 expression in U251, SK-MG-1 and the primary-cultured glioma cells derived from patients (NNS-10, NNS-11 and NNS-12). These results were then compared with those of the NY-ESO-1-expressing SaOS-2 cells (Fig. 2b). The expression of NY-ESO-1 was invariably induced by 5-aza-CdR in all glioma cells tested although the efficiency varied among cells; in particular, 10 μM 5-aza-CdR induced toxicity in SK-MG-1, leading to decreased NY-ESO-1 expression. Similar toxicity occurred in U87MG cells (Table II).
It is particularly important to ensure that CTA is not induced in normal cells, including normal brain cells, in order to prevent them from being targets of CTA-specific immune responses. We therefore treated human astrocytes, fibroblasts, smooth muscle cells and epidermal keratinocytes with 5-aza-CdR. In contrast to glioma cells, administration of the agent did not induce CTA expression in these cells (Fig. 2b and Table II).
To assess whether the induction of CTA mRNA is followed by the production of the appropriate protein, western blotting for NY-ESO-1 was performed using untreated and 5-aza-CdR-treated glioma cell lines. The SaOS-2 osteosarcoma cell line, which constitutively expresses NY-ESO-1 without 5-aza-CdR treatment, was used as a positive control. As shown in Figure 2c, 23-kDa bands corresponding to NY-ESO-1 were observed in all glioma cell lines treated with 5-aza-CdR but were absent in the untreated cells (all data not shown). This result suggested that the NY-ESO-1 expression was induced after 5-aza-CdR treatment at the protein level as well as the mRNA level.
Quantitative CpG island mapping with Pyrosequencing™
MSP experiments were performed to evaluate the methylation status of NY-ESO-1 in cultured glioma cells. We observed that the NY-ESO-1 is heavily methylated in glioma cells before 5-aza-CdR treatment and becomes hypomethylated following 5-aza-CdR exposure (Fig. 3a). To quantify the methylation of the CpG sites of NY-ESO-1, we employed a novel real-time DNA sequencing technology called Pyrosequencing™. This technology was originally developed for the analysis of single-base variations and enables the precise quantification of incorporated nucleotides at polymorphic positions. Treatment of the DNA with sodium bisulphite converts the epigenetic difference between methylated and unmethylated cytosine into a single-base variation of the C/T type. Therefore, Pyrosequencing is a very suitable tool for methylation analysis. Representative pyrograms for hypermethylated (untreated U251 cells) and hypomethylated (U251 cells following 5-aza-CdR exposure) are shown in Figure 3b. We identified the regions showing the largest differences in methylation and compared methylation levels of a small window (3 CpG sites) of NY-ESO-1. Mean methylation levels ± standard deviation (SD) were 86.3% ± 5.5%, 93.3% ± 0.6% and 93.3% ± 1.5%, for U251, T98 and NNS-10 glioma cells, respectively, while it was 30% ± 12% for SaOS2 cells constitutively expressing NY-ESO-1. Following 5-aza-CdR treatment, methylation levels were significantly decreased to 54% ± 9.1%, 68% ± 6.9% and 46.7% ± 5.7% for U251, T98 and NNS-10 (p < 0.05) (Fig. 3c). The MSP and Pyroseqencing data of other glioma cell lines and primary glioma cells were almost identical (data not shown). Taken together, this result is consistent with the hypothesis that 5-aza-CdR mediated NY-ESO-1 activation is a consequence of DNA demethylation.
Upregulation of HLA class I in glioma cells
Our microarray data (supplemental data and Discussion) indicated that HLA class I molecules can be upregulated by ∼3-fold (Table SII). Flow cytometric analysis confirmed that HLA class I expression was significantly increased in the 5-aza-CdR-treated glioma cells when compared to that in the untreated cells, indicating that 5-aza-CdR could affect the constitutive expression of HLA class I antigens in gliomas. The representative data of U251 cells are shown in Figure 4a. Combined with the analyses on the effect of 5-aza-CdR on NY-ESO-1 expression, our study suggests that 5-aza-CdR may have potential therapeutic implications in NY-ESO-1-specific immunotherapy for human gliomas.
Antigenicity of forcibly expressed NY-ESO-1 in glioma cells by 5-aza-CdR
To evaluate the antigenicity of forcibly expressed NY-ESO-1, HLA-A2-restricted NY-ESO-1-specific CTL lines were generated, and their cytotoxicity against 5-aza-CdR-treated glioma cells was tested. Cytotoxic activity was observed only in SaOS-2 osteosarcoma cells (NY-ESO-1+ and HLA-A2) and 5-aza-CdR-treated U251 glioma cells, depending on the E:T ratios (Fig. 4b). In contrast, untreated U251 cells and other glioma cells (NY-ESO-1-negative and HLA-A2-negative) were resistant to lysis. K562 cells were included in order to assess the degree of the natural-killer activity of the CTL cultures; this activity was found to be negligible. Cold target inhibition assays demonstrated that cytotoxicity against 5-aza-CdR-treated U251 cells was specifically inhibited in the presence of T2.A2 cells that were prepulsed with the cognate but not those that were prepulsed with an irrelevant peptide (Fig. 4c). This indicated that CTL lines recognized the NY-ESO-1 peptides that were processed and presented. These results together indicated that the amount of 5-aza-CdR-induced NY-ESO-1 in glioma cells was sufficient to render them sensitive to HLA-A2-restricted NY-ESO-1-specific CTLs in vitro.
Induction of NY-ESO-1 expression in intracerebral glioma by systemic administration of 5-aza-CdR
Previous evidence indicates that 5-aza-CdR can cross the BBB effectively, maintaining cytotoxic concentrations in the cerebrospinal fluid when administered via continuous intravenous infusion.46 To evaluate whether NY-ESO-1 expression could be induced in vivo after systemic delivery of 5-aza-CdR, we employed NOD/SCID mice transplanted intracerebrally with U251 glioma cells (NY-ESO-1 negative). After the mice were injected i.p. with a 0.2 mg/kg dose of 5-aza-CdR at 12 hr intervals for 4 days, the tumors were resected and subjected to RT-PCR analysis for NY-ESO-1. The results demonstrated that NY-ESO-1 expression could be detected in intracerebral gliomas i.p. treated with 5-aza-CdR (Fig. 5a).
Growth suppression of glioblastoma xenografts after the adoptive transfer of anti-NY-ESO-1 CTLs
Next, we addressed whether adoptively transferred NY-ESO-1-specific CTLs posses cytotoxic activity against gliomas in which NY-ESO-1 was induced in vivo by 5-aza-CdR. In this model, U251 tumor cells were stereotactically implanted in the forebrain of NOD/SCID animals. An intraperitoneal injection of 5-aza-CdR (or PBS) was initiated from the next day for 4 consecutive days at 12-hr intervals. On day 6 after the tumor inoculation, the mice were treated by stereotactic delivery of either NY-ESO-1-specific bulk CTLs or PBS. On day 21 after the tumor inoculation, tumor volumes were evaluated in half of the animals. Tumor growth was significantly delayed in the animals that were administered both 5-aza-CdR (i.p) and CTLs (Group IV), whereas relatively larger tumors were observed in the other 3 groups (Groups I, II and III, p < 0.05) (Fig. 5b). In addition, we measured the survival of the remaining half of the animals. The Kaplan–Meier survival curves of 4 groups are shown in Figure 5c. The survival time of mice treated with 5-aza-CdR and NY-ESO-1-specific CTLs was significantly greater than that of mice in Groups I, II and III (p = 0.0019, 0.0126 and 0.0132, respectively). Interestingly, 5-aza-CdR injection alone (Group III) exerted a beneficial antitumor effect in terms of tumor volumes and survival (p < 0.05 vs. Group I). As discussed later, the formation of enzyme-DNA adducts mediated by p53 induction may account for the efficacy and toxicity of 5-aza-CdR in vivo.
The principal findings of this study are that in spite of relatively low frequency of CTA expression in gliomas, the DNA demethylating agent 5-aza-CdR remarkably induced the expression of CTAs, including NY-ESO-1-one of the most immunogenic CTAs-in glioma cells but not in normal human cells. The de novo expressed NY-ESO-1 was effectively recognized by the specific CTL lines both in vitro and in glioma orthotopic xenografts.
Expression of CTAs in human gliomas
A German group investigated the expression of 7 CTAs (SSX-1, SSX-2, SSX-4, SCP-1, TS85, NY-ESO-1 and MAGE-3) in 50 gliomas by RT-PCR. They demonstrated that SCP-1 was most frequently positive (40%), followed by SSX-4 (27%) and MAGE-3 (7%).35 However, the expression frequency of CTAs in gliomas remains unclear, particularly when focusing on different ethnic groups. For example, NY-ESO-1 and LAGE-1 are expressed at much lower frequencies in lung cancer in the Japanese than in Caucasians; NY-ESO-1 is expressed in 2% of the Japanese versus 17 or 20% in Caucasians, and LAGE-1 is expressed in 9% of the Japanese versus 33% expression in Caucasians.39 Liu et al., have reported MAGE-1 expression in approximately 40% of glioblastoma primary cell lines and in established glioma cell lines, including U87MG, which is inconsistent with our results.47 Although the reasons for the inconsistency remain unclear, the differences in ethnic groups and culture conditions (e.g., passage and confluency) might be responsible. Nevertheless, overall, we found that gliomas have a very low frequency of CTA gene expression, such as NY-ESO-1, indicating that brain tumors are considerably unsuitable for CTA-based immunotherapy. To overcome this limitation, novel CTA induction strategies are required to evoke strong immune responses against gliomas.
CTA as targets of demethylation
A cascade of biochemical events for gene silencing is triggered by CpG island methylation that involves DNA methyltransferase activity, which in turn attracts histone deacetylases and histone methylases that eventually modify histones into a silenced chromatin state.48 The agent 5-aza-CdR has been shown to interrupt this silencing cascade effectively by binding covalently to DNA methyltransferase and inhibiting its enzymatic activity. Numerous studies have shown the ability of 5-aza-CdR to reactivate the transcription of several tumor suppressor genes (i.e. p16 and MGMT) in human tumors.28, 29 Previous evidence has clearly defined the regulatory role of DNA methylation in the constitutive expression of CTAs in melanomas and renal cell carcinomas and has demonstrated that in vitro treatment with 5-aza-CdR induces their expression in neoplastic cells. Therefore, CTAs are intriguing targets for demethylation. Here, we studied a large panel of CTAs and showed that 5-aza-CdR reactivated the expression of a variety of CTAs in human gliomas. This result is consistent with the report of Liu et al., showing that 5-aza-CdR induced the mRNA expression of MAGE-1.47 Then, we determined whether 5-aza-CdR-mediated NY-ESO-1 activation is a consequence of promoter DNA methylation by using MSP and quantitative Pyrosequencing. Although MSP is very sensitive and easy to use, it does not provide precise information about the methylation status of single CpG sites. Recently, Pyrosequencing technology was developed for the analysis of single-base variations. An indirect bioluminometric assay quantitatively measures the amount of pyrophosphate (Ppi) that is released from each incorporated dNTP. Through an enzymatic cascade, the release is converted into a light signal that is directly proportional to the amount of incorporated dNTP, appearing as peaks in a Pyrogram. Employing this technology, we first showed that only a small region of the NY-ESO-1 CpG island provides relevant information for differential methylation analysis.
Expression profiling and gene ontology analysis after 5-aza-CdR treatment
To identify alterations in gene expression after 5-aza-CdR treatment, we conducted microarray experiments (supplementary data). Once the analysis was completed, we narrowed down the number of potential targets by selecting only those genes whose expression changed more than 2-fold in 2 independent RNA preparations. A total of 65 genes that fulfilled our criteria were upregulated, and 24 genes were downregulated following 5-aza-CdR treatment (Table SII). To identify the biological processes significantly involved in the drug effect, genes considered differentially expressed between the treated and control glioma cells were processed through the Gene Ontology (GO) program. The majority of the differentially expressed genes were related to biological process categories such as apoptosis (programmed cell death), cell proliferation, immune system process and tissue development (Table SIII). These categorized GO terms supported the fact that DNA methylation is associated with various biological epigenetic processes, including cell differentiation, development and oncogenic transformation. For the 11 upregulated genes underlined in Supplementary Table SII, we confirmed the microarray data using semiquantitative RT-PCR with the primers listed in Supplementary Table SIV on the same glioma cell line (U251) and 3 others (AO2, T98 and SKMG1). As shown in Supplementary Figure S1, we were able to show marked up-regulation of the tissue inhibitor of metalloprotease (TIMP) gene, which has been found to be silenced by aberrant promoter hypermethylation in other tumor types,49 thus validating our screening procedures. Interestingly, p53, Gadd45, NF-κB and caspase 4 genes were activated by 5-aza-CdR, although the dose of 5-aza-CdR in this study (1 μM) did not affect the cell viability and morphology in glioma cells (data not shown). Recently, Kim et al., attempted to identify genes silenced epigenetically in malignant gliomas by using a comprehensive and intense microarray technique coupled with the inhibition of DNA methylation and histone deacetylation.50 Although they validated the reactivation of the MAGE genes in their micoarray screen, they mainly focused on novel targets harbouring the CpG island promoter. In addition to the role of 5-aza-CdR in the activation of epigenetically silenced genes, an important biological activity of this agent is the formation of enzyme-DNA adducts.51 Karpf et al., reported that the formation of covalent enzyme-DNA adducts and cellular toxicity resulted from the activation of p53 as a cellular response to DNA damage.52
5-Aza-CdR as a potent immunostimulator
Our microarray data (Supplementary data) and the flow cytometric analysis showed that HLA class I expression was significantly increased in the 5-aza-CdR-treated glioma cells. In addition, the efficient recognition of 5-aza-CdR-treated U251 glioma cells by the HLA-A2 restricted NY-ESO-1 specific CTL lines demonstrated that de novo synthesized NY-ESO-1 antigen is functionally processed and presented. Thus, CTL-mediated lysis of glioma cells induced by 5-aza-CdR appears to represent a direct consequence of immunogenic peptides derived from de novo expressed NY-ESO-1 and loaded onto upregulated HLA class I molecules. Our in vivo study suggested that systemic administration of 5-aza-CdR may be useful in reverting the CTA-negative phenotype of gliomas through the Blood-Brain-Barrier. This evidence strongly identifies 5-aza-CdR as a potential pharmacological agent in designing and establishing new therapeutic strategies in combination with CTA-based immunotherapeutic approaches for glioma patients.
The authors thank Mr. Shigeru Saito (Infocom Corporation, Tokyo, Japan) for bioinformatics assistance in Gene Ontology analysis. They also thank Mr. Mitsuo Kihira (Division of Pathology, Kariya Toyota General Hospital, Aichi, Japan) for assistance in the histological study.