Decitabine immunosensitizes human gliomas to NY-ESO-1 specific T lymphocyte targeting through the Fas/Fas Ligand pathway

Five cell lines, DBTRG05-MG, SNB-19, U-251MG, U-373MG and T98G derived human glioblastoma cell lines were kindly provided by Dr. Carol Kruse (UCLA Department of Neurosurgery, Los Angeles, CA). They were maintained in DMEM (Mediatech, Inc., Herndon, Virginia) supplemented with 10% FBS (Gemini Bio-Products, West Sacramento, California), 1% (v/v) penicillin, streptomycin, and amphotericin B (Mediatech Cellgro, Manassas, Virginia) and kept in an atmosphere of 5% CO₂ at 37°C. Normal human astrocytes (NHA) were kindly provided by Dr. Russ Pieper (UCSF Department of Neurosurgery, San Francisco, CA). A melanoma cell line, 624.38, was kindly provided by Dr. Steve Rosenberg (NIH/NCI). Both NHA and 624.38 cells were placed into identical culture conditions to glioblastoma cell lines.

Primary tumor cell cultures were derived from four patients with glioblastomas who had undergone surgical resection at the UCLA Medical Center. These tissues were cultured by digesting homogenized tumors in a collagenase DNAse mixture overnight as previously described [5, 29]. The digested tumor samples were filtered through a mesh and the cells placed into complete medium as described earlier. The cells were incubated in 5% CO₂ at 37°C. Human brain tissue samples were also obtained from patients who had undergone resection at the UCLA Medical Center, CA. All patients provided written informed consent for these procedures.

Peripheral blood was obtained from healthy human volunteers provided by the Division of Hematology and Oncology at UCLA Medical Center, Los Angeles. Peripheral blood mononuclear cells (PBMC) were isolated by Ficoll gradient centrifugation as previously described [5]. Written informed consent and institutional IRB approval was obtained for all studies involving human bloods and tissues.

Decitabine (5-aza-2'-deoxycytidine) was generously supplied by Eisai Pharmaceuticals (Woodcliff Lake, New Jersey). A 10 μM stock solution of it in DMSO was stored at -80°C.

Cells were plated overnight at 10⁶cells/ml at 37°C in a 5% CO₂ incubator. The following day, the cell culture medium was replaced with either fresh medium or that supplemented with 1 μM decitabine. The cells were treated again the following day in new cell culture medium. At the end of treatment, the medium was replaced with fresh medium without decitabine, and the cells were cultured for an additional 48 hours before utilized in subsequent assays.

Total RNA was isolated with the RNeasy Mini kit (Qiagen, Valencia, California) according to manufacturer's protocol. 3 × 10⁶ human glioblastoma cancer cell lines or 25 mg of tumor tissue samples were used. cDNA was synthesized from 1 μg total RNA by using Omniscript Reverse Transcription kit (Qiagen), again according to the manufacturer's protocol. PCR was facilitated by using the Accuprime GC-rich kit (Invitrogen, Carlsbad, California). In each PCR, 2 μl cDNA was used in a 35-μl reaction volume containing 10 μM sense and antisense primers, 5 μL 5× PCR buffer (A for GC-rich genomic DNA target or B for non-GC-rich genomic DNA target), and 1 μl of Accuprime GC-Rich DNA polymerase (Invitrogen). The PCR primers were as follows: NY-ESO-1 forward primer: 5'-CATCACGGATCCATGCAGGCCGAAGGCCGG-3', reverse primer: 5'-ACCCGGGGTACCGCGCCTCTGCCCTGAGGG-3', GAPDH forward primer: 5'-GAAGGTGAAGGTCGGAGT-3', reverse primer: 5'-GAAGATGGTGATGGGATTTC-3' (Invitrogen). The PCR cycling parameters were as follows: denaturation at 95°C for 30 s, primer annealing at 60°C for 30 s, and 40 cycles of extension at 72°C for 60 s. PCR cycling was preceded by an initial denaturation at 95°C for 3 min, followed by final extension at 72°C for 10 min. The PCR products were electrophoresed on 1% agarose gels and analyzed after staining with ethidium bromide. Densitometry analysis was performed using the QuantityOne program (BioRad, Hercules, CA).

Quantitative RT-PCR was performed with the LightCycler real-time RT-PCR system (Roche, Mannheim, Germany), using the LightCycler SYBR green mastermix (Roche). Primers specific for NY-ESO-1 were as follows: forward primer: 5'-TGTCCGGCAACATACTGACT-3', reverse primer: 5'-ACTGCGTGATCCACA TCAAC-3'. GAPDH forward primer: 5'-AGCCACATCGCTCAGACAC-3', reverse primer: 5'-CGCCCAATACGACCAAATC-3' (Invitrogen). Each sample was amplified as follows: 1 cycle at 95°C for 10 min, 40 cycles at 95°C for 15 s, and 60°C for 1 min. Triplicate samples were tested.

Total RNA was extracted from decitabine treated and untreated U-251 MG cells using the RNeasy mini kit (Qiagen). cDNA was generated, quantified and hybridized to U133 Plus 2.0 arrays at the UCLA DNA Microarray Facility using standard Affymetrix protocols. CEL files were normalized using GeneSpring GX 11.5.1 software (Agilent Technologies) with LiWong. To evaluate the baseline expression of NY-ESO-1 in heterogeneous human brain tumor tissues, we analyzed the relative expression of NY-ESO-1 using microarray gene expression profiling in over 300 human brain tumor samples. The relative expression of NY-ESO-1 was normalized and tested. Dots plotted underneath the black line are considered to have at or under background relative expression levels.

Cells (10⁵) obtained from decitabine-treated and untreated U-251 MG cells were used for flow cytometric analysis. Cells were washed twice with FACS staining buffer and stained with antibodies to HLA-A, B, C (clone DX17) and HLA-A2 (clone BB7.2) (BD Pharmingen, San Diego, California). A Becton Dickinson LSRII was used for acquisition and the acquired data were analyzed using FlowJo (TreeStar, Ashland, Oregon). Gates were set based on singly-stained isotype-control antibodies (data not shown).

Six well cluster plates were coated with Retronectin (RN) (Takara, Madison, Wisconsin) at a concentration of 10 μg/ml overnight at 4°C. The next day, blocking buffer (PBS containing 2% Human Serum Albumin) was added for 30 min and washed. Supernates were retrieved from a stable PG13-based retroviral producer cell line encoding a HLA-A*0201-restricted NY-ESO_157-165 specific T cell receptor (TCR) generated in the MSGV1 retroviral vector backbone, as described [30]. The PG13-based NY-ESO-1 stable retroviral packaging cell line was obtained from Dr. Paul Robbins (Surgery Branch, NCI/NIH). Anti-CD3 (clone OKT-3; 50 ng/ml) was used to stimulate human PBMCs for 48 hours in X-VIVO 15 medium supplemented with 5% human AB serum in the presence of 300 IU IL-2/ml prior to spin-infection. Stimulated PBMCs were transduced twice as described [15, 30]. Transduced T cells were maintained in X-VIVO 15 medium supplemented 5% human AB serum and 300 IU IL-2/ml. The cells were expanded for 3 days in the presence of IL-2 and then rested for 2 days in the absence of IL-2.

Transduction efficiency was evaluated 48 hr post-transduction by NY-ESO-1 T cell receptor staining of the CD3⁺CD8⁺ gated population using antihuman antibodies to CD3-PerCP (clone SK7; BD Biosciences, San Diego, CA), CD8-FITC (clone SFCI21Thy2D3; Beckman Coulter, Brea, CA), and TCR Vβ13.1-PE (clone IMMU 222; Beckman Coulter). Detection of the NY-ESO-1 peptide-MHC complex was accomplished by NY-ESO-1 tetramer staining as described [31].

NY-ESO-1 TCR-transduced PBMCs and untransduced PBMCs were harvested and resuspended at a concentration of 1 × 10⁷ cells/ml. Treated and untreated HLA-0201⁺ T98G glioma cells were harvested using trypsin and reconstituted to 1 × 10⁷ cells/ml. 1 × 10⁶ PBMCs were co-incubated with target T98G cells in 96-well round-bottomed plates at a 1:1 ratio with total medium volume of 200 μl. Golgi plug (1×) and 5 μl of CD107A-AF647 (clone H4A3) (BD Biosciences) were added into each well and incubated for 6 hr at 37°C. Cells were washed twice with FACS staining buffer (PBS + 2% FBS) and stained with surface antibodies to CD4-PE (clone RPA-T4; BD Biosciences), CD8-PacBlue (clone RPA-T8; BD Biosciences), and CD3-PE-Cy5 (clone UCHT1; BD Biosciences) on ice for 25 min. Cells were washed, fixed with Fixation Buffer (eBioscience, San Diego, CA), and kept at 4°C until analysis. Data were collected using a Becton Dickinson LSRII and analyzed with FlowJo software (TreeStar).

To determine cytokine concentrations, supernates were collected from the Golgi-plug free overnight co-cultures. Cell-free supernates were analyzed for interferon-γ using an ELISA kit (eBiosciences) or for simultaneous levels of IFN-γ, TNF-α, IL-2, IL-4, IL-5, and IL-10 using a cytokine bead array (CBA) system (BD Biosciences), following the manufacturer's instructions. Co-cultures at 1:1 (E:T) ratios were placed into complete medium (X-VIVO 15 with 5% human AB serum) in 96-well round-bottomed plates and incubated overnight at 37°C. All samples were in triplicate and supernates were harvested for cytokine detection. IFN-γ ELISAs were performed according to the manufacturer's specification (ELISA Ready-set-go, eBioscience). Color development was stopped after 2-5 minutes. The colorimetric density of each well was measured at 450 nm by a plate reader, and the final concentration of each sample was calculated (pg/ml) based on the standard curve. Triplicate assays were performed for each case and the results were expressed as the mean of 3 separate assays.

Data acquisition for CBA experiments was performed with the FCAP Array software (BD). Once the FCAP Array experiment was created, the files were exported to the BD FACSArray software for sample acquisition and analysis following manufacturer's protocol.

Apoptosis was induced in T98G glioma cells or 624.38 melanoma cells by the addition of an agonistic anti-CD95 antibody (clone CH-11; Medical and Biological Laboratories). Cells were analyzed for Fas-expression using a CD95-PE mAb (clone DX2) (BD Pharmigen). On the day of harvest, the target cells were cultured in DMEM with FBS containing 200 ng/ml of CH-11 for 24 hr. Cells were washed and resuspended in 1× binding buffer (Invitrogen). Cells were then stained with FITC-conjugated Annexin V (Invitrogen) and the vital dye, propidium iodide (PI, Invitrogen), for 20 min at 37°C. Cells negative for both PI and Annexin V staining were considered live cells; early apoptotic cells were PI-negative, Annexin V-positive; and late apoptotic/dead cells were PI-positive, Annexin V-positive. Each specimen was analyzed in triplicate, and the results shown were the mean ± SE of 3 assays.

PBMCs transduced with the NY-ESO-1 TCR were co-cultured with tumor target cells and blocked with an antagonistic anti-CD95 antibodies (clone ZB4; Medical and Biological Laboratories). Treated and untreated HLA-0201⁺ T98G glioma cells were harvested using trypsin and reconstituted to 4 × 10⁵ cells/ml and blocked with antagonistic anti-CD95 antibodies for 1 hr on ice. Cells were washed and reconstituted to 1 × 10⁷ cells/ml in complete medium. Transduced PBMCs were plated with target cells in 96-well round-bottomed plates at a 1:1 ratio with total medium volume of 200 μl. Golgi plug (1×) and 5 μl of CD107A-AF647 (clone H4A3) (BD Biosciences) were added into each well and incubated for 6 hr at 37°C. Cells were washed twice with FACS staining buffer and stained with surface antibodies to CD4-PE, CD8-PacBlue, and CD3-PE-Cy5 on ice for 25 min. Cells were fixed with Fixation Buffer and kept at 4°C until analysis. Data were collected using a Becton Dickinson LSRII and analyzed using FlowJo software (TreeStar).

We previously demonstrated that many melanoma associated antigens (MAA) are expressed by glioma cells, but at distinctly lower levels compared with melanoma cells [32]. To identify epigenetically silenced genes in glioma cells, we treated cells with the DNA methyltransferase inhibitor, decitabine, and then performed global gene expression profiling. Many CTA and major histocompatibility complex (MHC class I) antigens can exhibit methylation in their promoters, thus influencing their protein expression at an epigenetic level. As shown in Figure 1A, an unbiased microarray screening ranked a significant number of CTA as being highly up-regulated after decitabine treatment. These data included a 58-376 fold increase in NY-ESO-1 and other related CTA (Table 1). Multiple genes from the MHC and death receptor family were also upregulated (Additional File 1). To confirm changes in the protein expression of MHC, we stained cells with monoclonal antibodies to pan class I (HLA-A, B, C) as well as to HLA-A2 specifically. Decitabine treatment resulted in increased surface expression to MHC-class I molecules and to HLA-A2 (Figure 1B).

To confirm our microarray results, conventional RT-PCR and quantitative RT-PCR were used to assess NY-ESO-1 expression in established glioma cell lines and primary glioblastoma cell explants placed into culture immediately after surgical resection (Figure 2). Multiple glioma cell lines allowed us to assess whether the induction of NY-ESO-1 expression was a consistent feature of decitabine treatment. Results showed a uniform up-regulation of NY-ESO-1 in all decitabine-treated cell lines, with some variation in levels of expression (Figure 2A). Decitabine-treated primary cell lines also showed uniform up-regulation of NY-ESO-1 expression (Figure 2B). However, such up-regulation of NY-ESO-1 did not occur in non-malignant, normal human astrocytes (NHA), alleviating potential concerns that systemic administration of decitabine will up-regulate CTA on normal brain cells (Figure 2C). An NY-ESO-1⁺HLA-0201⁺ melanoma cell line (624.38) was used as a positive control. The high endogenous expression of NY-ESO-1 was not changed with decitabine treatment (Figure 2C). To quantify the induction of NY-ESO-1 after decitabine treatment, real-time PCR was performed, which correlated with the results obtained with conventional RT-PCR (Figure 2D). Compared to the untreated control cells, the decitabine-treated T98G and glioblastoma #1 cells showed average log fold increases in the gene expression of NY-ESO-1 (3155.12 and 544.29, respectively) both of which were statistically significant (p = 0.0001 and 0.0075).

To evaluate the baseline expression of NY-ESO-1 in heterogeneous human brain tumor tissues, a large panel of normal tissues and brain tumors of varying grades were subjected to global gene expression profiling using Affymetrix U133 2.0 chips. NY-ESO-1 was not expressed in lower-grade gliomas (WHO Grades II-III). However, significantly elevated expression was detectable in a small percentage of medulloblastoma and glioblastoma (WHO Grade IV) tumor specimens. Such expression was comparable to the expression of NY-ESO-1 observed in normal testes, the only non-malignant, postnatal tissue that normally expresses this CTA (Figure 3). These data clarify the frequency and degree of expression of NY-ESO-1 in heterogeneous patient samples.

To evaluate whether up-regulated expression of NY-ESO-1 and MHC I sensitized these tumor cells to T cell recognition, we expressed an NY-ESO-1 specific T-cell receptor clone 1G4 in normal human PBMCs utilizing a retroviral transduction system [30]. Within the transduced CD3⁺CD8⁺ T-cell population of PBMCs, 56% of cells expressed TCRVβ13.1, a TCR known to be expressed by the clone 1G4-α95:LY NY-ESO-1 TCR [30] (Additional File 2). Untransduced cells showed only small frequencies of a TCRVβ13.1⁺ T cell population. Tetramer staining revealed that nearly 50% of the CD3⁺CD8⁺ population was NY-ESO-1 specific (Figure 4A). Tetramer staining also showed that approximately 45% of the CD3⁺CD4⁺ population was NY-ESO-1 specific.

To evaluate the functional recognition of glioma cells by NY-ESO-1 specific T cells, the cells were co-cultured and then stained for intracellular expression of CD107A (LAMP-1). The cell surface mobilization of CD107A has been directly linked to CTL lytic granule release and target cell death [33]. In a representative study that has been repeated at least 3 times, NY-ESO-1 specific CD8⁺ T cells, gated from the CD3⁺CD8⁺ population, exhibited a 7% or a 68% expression of CD107A when co-cultured with control or decitabine-treated T98 glioma cells, respectively (Figure 4B). NY-ESO-1 specific CD4⁺ T cells, gated from the CD3⁺CD4⁺ population, did not respond to the co-cultures (data not shown). NY-ESO-1 specific T cells co-cultured with an HLA-A*0201⁺, NY-ESO-1⁺ melanoma line (624.38) exhibited a 65% expression of CD107A (data not shown). To assess the release of cytokines by NY-ESO-1 specific T cells, we used the cell-free supernates of overnight co-cultures for ELISA and CBA. As shown in Figure 5A, significantly elevated concentrations of IFN-γ (1133.9 pg/ml) were detected when NY-ESO-1 specific T cells were co-cultured with decitabine-treated T98 glioma cells, whereas, co-culture with untreated glioma cells was below the level of detection. Co-culture of untransduced PBMCs with control or decitabine-treated glioma cells also did not elicit the release of detectable IFN-γ. Using cytometric bead arrays, we were able to detect multiple Th1/Th2 cytokines simultaneously secreted by NY-ESO-1 specific T cells. Significant differences in the concentrations of IFN-γ, TNF-α, and IL-5 were observed only when NY-ESO-1 specific T cells were co-cultured with decitabine-treated T98 glioma cells (Figure 5B; p < 0.0001). Small differences were also seen with IL-2 and IL-4, but not IL-10 (data not shown). Thus, treatment of T98 glioma cells with decitabine results in significantly enhanced NY-ESO-1 specific T cell recognition and secretion of Th1-type cytokines along with IL-5.

To test whether decitabine treatment sensitized glioma cells to distinct immune-mediated death receptor pathways, we assessed expression of TNFR, TRAIL-DR4, TRAIL-DR5, and Fas on T98 glioma cells by flow cytometric analyses. No changes in the expression of TNFR, TRAIL-DR4, or TRAIL-DR5 were observed after decitabine treatment (data not shown). Small, but detectable increases in the mean fluorescence intensities (MFI) of Fas (CD95) were observed after decitabine treatment (Figure 6A). Furthermore, addition of an agonistic Fas mAb to human T98 glioma cells enhanced tumor cell death when pre-treated with decitabine (Figure 6B). Staining of T98G glioma cells with Annexin V and propidium iodide (PI) indicated the average percentages of late apoptotic cells at 17%, 26%, 20%, and 51% in untreated, untreated with Fas mAb, decitabine-treated alone, and decitabine-treated with Fas mAb, respectively (Figure 6B). One-way ANOVA statistical testing showed a significant difference between the control and Fas-antibody treated groups (p < 0.01). There was also a significant difference observed within untreated and decitabine-treated groups with the addition of the agonistic Fas mAb (*p = 0.0215 and **p = 0.0049). Such effects on Fas-mediated apoptosis were consistently observed at timepoints as early as 4-6 hours and as late as 4 days (data not shown).

Because decitabine pre-treatment sensitized glioma cells to Fas mAb mediated cell death with an agonistic mAb, we also tested whether such death receptor pathways were functionally important for NY-ESO-1 specific T cell recognition and killing of glioma cells. Untreated T98G glioma cells, blocked with a Fas antagonistic antibody, showed no difference in CD107A expression compared to untreated T98G glioma cells co-cultured with NY-ESO-1 specific T cells (Figure 6C). However, decitabine-treated T98G glioma cells, blocked with a Fas antagonistic antibody, showed dramatic decreases in CD107A expression levels compared to decitabine-treated T98G glioma cells co-cultured with NY-ESO-1 specific T cells (Figure 6D). The NY-ESO-1 specific T cells alone (negative control) showed minimal release of CD107A expression. To address the effect of blocking the Fas pathway on cytokine release, supernatants derived from NY-ESO-specific T cells co-cultured with T98G cells were tested in cytokine bead arrays. Addition of the antagonistic Fas mAb to decitabine-treated cells largely inhibited the entire induction of Th1-type cytokine release (IFN-γ and TNF-α) by NY-ESO-1 specific T lymphocytes (Figure 6E). IL-10 concentrations appeared to decrease but were not as prominent as the other Th1/Th2 cytokines (data not shown). IL-5 concentrations were almost completely inhibited as well.