Purpose: Successful immunotherapies for IDHmut gliomas require better knowledge of T-cell target antigens. Here, we elucidated their antigen repertoire recognized by spontaneous T-cell responses using an unbiased proteomic approach.

Experimental Design: Protein fractionations of tissue lysates from IDHmut gliomas (n = 4) were performed. Fractions were tested by IFNγ ELISpot assay for recognition through patients' T cells. Proteins of immunogenic fractions were identified by mass spectrometry and validated by in silico-predicted synthetic long peptides in patients of origin, additional IDHmut glioma patients (n = 16), and healthy donors (n = 13). mRNA and protein expression of immunogenic antigens was analyzed in tumor tissues and IDHmut glioma stem-like cells (GSC). HLA-A*02–restricted T-cell epitopes were functionally determined by short peptides and numbers of antigen-specific T cells by HLA-peptide tetramer analysis.

Results: A total of 2,897 proteins were identified in immunogenic tumor fractions. Based on a thorough filter process, 79 proteins were selected as potential T-cell antigens. Twenty-six of these were recognized by the patients’ T cells, and five of them (CRKII, CFL1, CNTN1, NME2, and TKT) in up to 56% unrelated IDHmut glioma patients. Most immunogenic tumor-associated antigens (TAA) were expressed in IDHmut gliomas and GSCs, while being almost absent in normal brain tissues. Finally, we identified HLA-A*02–restricted epitopes for CRKII, NME2, and TKT that were recognized by up to 2.82% of antigen-specific peripheral cytotoxic T cells in IDHmut glioma patients.

Conclusions: By analyzing the repertoire of T-cell target antigens in IDHmut glioma patients, we identified five novel immunogenic TAAs and confirmed their expression on IDHmut tumors and GSCs. Clin Cancer Res; 24(12); 2951–62. ©2018 AACR.

Translational Relevance

IDH1-mutant lower-grade gliomas (LGG) are characterized by small numbers of neoantigens due to a low mutational load and therefore are regarded as immunologically “cold” tumors. With the help of an unbiased proteomic-based approach, we elucidated the repertoire of immunogenic, nonmutated T-cell target antigens in LGG. A thorough characterization of these antigens revealed a considerable expression in IDH1-mutant glioma stem-like cells, overexpression in tumor tissues, and a tumor-specific and frequent immunogenicity in up to 56% of IDH1-mutant glioma patients. Notably, frequencies of antigen-specific T cells in the peripheral blood reached up to 2.8% of all T cells. Our data demonstrate frequent spontaneous T-cell responses against tumor-associated antigens, suggesting their use as suitable targets for future immunotherapy of IDH1-mutant gliomas.

Gliomas are classified by the World Health Organization (WHO) according to the degree of malignancy from WHO°I–IV (1). Lower-grade gliomas (LGG) referring to low-grade WHO°II and anaplastic WHO°III gliomas are characterized by an invasive growth, the ability to progress to higher-grade lesions up to secondary glioblastomas (sGBM, WHO°IV), and a common missense mutation in the isocitrate dehydrogenase (IDH)-1, which is rarely found in primary glioblastomas (pGBM; refs. 1, 2). The IDH mutation (IDHmut) increases the enzymatic conversion of α-ketoglutarate into the oncometabolite 2-hydroxyglutarate driving a glioma CpG island-methylated phenotype (G-CIMP) with a profound impact on biology, protein expression, and survival (2–4). In addition to the IDH mutation, occurrence of a codeletion on the chromosome arms 1p and 19q led to a further subclassification into astrocytoma (IDHmut, 1p/19q no-codeletion) and oligodendroglioma (IDHmut, 1p/19q codeletion; ref. 1).

Despite some therapeutic advances in LGGs, resistance to conventional therapies and their diffuse nature still limits therapeutic success (5). As a potential novel treatment option, the field of cancer immunotherapy is showing first success by the FDA approval of immunomodulatory checkpoint inhibitors yet in other tumor types (6, 7). However, even in gliomas, immunotherapeutic approaches gain more attention, since the traditional perception of the brain has changed from an immune-privileged to an immune-active organ (8). This enables the application of novel treatments for gliomas such as adoptive T-cell transfer (9) as well as vaccinations, utilizing tumor-associated antigens (TAA), mutated tumor-specific antigens (TSA), or even whole tumor lysates (10). However, in contrast to other tumor entities, in LGGs, the number of TSAs is low due to a very low somatic mutation rate (11). Even for TAAs, a comprehensive repertoire has not been established because so far only few TAAs which have been characterized in other neoplasms were found to be expressed on glioma cells (12). Hence, vaccinations with defined TAAs for the treatment of gliomas are still rare (13, 14).

Previous approaches to identify TAAs mainly examined IDH wild-type (IDHwt) rather than IDHmut gliomas and applied indirect methods such as mRNA expression analysis, complementary DNA libraries, or determined MHC-bound peptides yet without addressing their endogenous immunogenicity (15–17). The importance of endogenous anti-TAA T-cell responses, however, has been underlined by a recent project of the NCI prioritizing immunogenicity of antigens as one of the most important selection criteria for an ideal target (18).

In the present study, we made use of an unbiased proteomic-based approach to unravel the T-cell target antigen repertoire of IDHmut gliomas and thereby identified CRKII, CFL1, CNTN1, NME2, and TKT as novel immunogenic TAAs triggering spontaneous T-cell responses in up to 56% of IDHmut glioma patients. Furthermore, we defined corresponding HLA-A*02–restricted epitopes recognized by up to 2.82% antigen-specific peripheral T cells.

Patients

To identify T-cell target antigens in LGGs, we obtained specimens from patients undergoing surgical resection at the Department of Neurosurgery, Heidelberg University Hospital (Supplementary Table S1). Snap-frozen tumor tissues were stored at −80°C. Tumor cell content ≥60% and IDH1(R132H) mutation status of all samples were confirmed by experienced neuropathologists (A. von Deimling and D. Reuss). Use of patient material was approved by the Heidelberg Medical Faculty Review Board in accordance with the Declaration of Helsinki. Written informed consent was obtained from all patients.

Tissue lysis

Small tumor pieces were mechanically dissected using an Ultra-Turrax (IKA) and lysed in 2 mL of lysis buffer [5 mol/L urea (#X999.3); 2 mol/L thiourea (#HN37.2); 10% glycerol (#3783.1); 50 mmol/L Tris-HCL (pH 8; #9090.2, all Carl Roth); 2% n-Octylglucoside (#10634425001); 2.5% SB3-10 (#D4266; both Sigma-Aldrich); 1 mmol/L complete protease inhibitor (#11836153001; Roche Diagnostics)]. For desalting, the buffer was subsequently exchanged by PD-10 Sephadex G-25 columns (#17085101; GE Healthcare) and eluted in 3.5 mL starting buffer (#A51675; Beckman Coulter). The Micro BCA Protein Assay Kit (#23235; Thermo Fisher Scientific) was used to determine the protein concentration.

PF2D liquid chromatography

To identify T-cell target antigens, we conducted a two-dimensional tumor proteome fractionation by injecting 2.5 mg of tumor lysates into the ProteomeLab Two-dimensional Protein Fractionation System (PF2D; Beckman Coulter) as described previously (ref. 19; Fig. 1A). In the first dimension (1D), proteins were separated by chromatofocusing. 1D protein fractions were further separated in the second dimension (2D) by reversed phase high-pressure liquid chromatography. Selected fractions were loaded on dendritic cells (DC) generated from blood-derived monocytes of corresponding patients as described (19) and used for immunological testings (IFNγ ELISpot) with purified T cells from the same patients (Fig. 1B and C). Identification of proteins in recognized fractions was conducted by liquid chromatography electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS).

Figure 1.

A, Workflow of the combined PF2D and IFNγ ELISpot method to identify immunogenic T-cell targets. Tumor lysates were injected in the first dimension and separated by chromatofocusing. Fractions were collected in a 96-well plate in intervals of 0.3 pH within the range of pH 8.5– 4.0. Based on the first-dimension chromatogram, fractions were selected for immunologic testings in an autologous setting by IFNγ ELISpot assay. Immunogenic fractions were reinjected in the second dimension, separated by reversed phase chromatography, fractionated, and tested by IFNγ ELISpot. Immunogenic second-dimension fractions were further processed and analyzed by mass spectrometry to identify the proteome content. Mass spectrometry–identified proteins passed a comprehensive selection process, and the immunogenicity was validated by means of long (50-AA) synthetic peptides in LGG patients and healthy donors. Identified targets were further characterized by flow cytometry, in situ immunohistochemical staining, and bioinformatics approaches using RNAseq TCGA data. Graphic representation of IFNγ spot numbers by T cells, cocultivated with autologous tumor lysates of (B) first-dimension and corresponding (C) second-dimension PF2D tumor fractions of one representative patient (NCH1150). F, fractions; TC, T cells. *, P < 0.05; **, P < 0.01; ***, P < 0.001, compared with PBL-negative control (solid line).

Figure 1.

A, Workflow of the combined PF2D and IFNγ ELISpot method to identify immunogenic T-cell targets. Tumor lysates were injected in the first dimension and separated by chromatofocusing. Fractions were collected in a 96-well plate in intervals of 0.3 pH within the range of pH 8.5– 4.0. Based on the first-dimension chromatogram, fractions were selected for immunologic testings in an autologous setting by IFNγ ELISpot assay. Immunogenic fractions were reinjected in the second dimension, separated by reversed phase chromatography, fractionated, and tested by IFNγ ELISpot. Immunogenic second-dimension fractions were further processed and analyzed by mass spectrometry to identify the proteome content. Mass spectrometry–identified proteins passed a comprehensive selection process, and the immunogenicity was validated by means of long (50-AA) synthetic peptides in LGG patients and healthy donors. Identified targets were further characterized by flow cytometry, in situ immunohistochemical staining, and bioinformatics approaches using RNAseq TCGA data. Graphic representation of IFNγ spot numbers by T cells, cocultivated with autologous tumor lysates of (B) first-dimension and corresponding (C) second-dimension PF2D tumor fractions of one representative patient (NCH1150). F, fractions; TC, T cells. *, P < 0.05; **, P < 0.01; ***, P < 0.001, compared with PBL-negative control (solid line).

Close modal

Data analysis and protein selection process

LC-ESI-MS/MS spectra were searched against the SwissProt database (http://www.uniprot.org) and provided as Mascot Search Results (http://www.matrixscience.com) as described (20). A protein was considered as identified if at least two corresponding peptides were detected. For diminishing false-positive hits, criteria were as following: protein sequence coverage > 10% and number of peptides matched (n > 5; Fig. 2A). Frequent contaminations such as keratin (1, 2, 9, 10, 21), hornerin (22), and dermcidin caused by human skin, dandruff, or sweet glands were excluded. All remaining proteins were characterized in terms of function- and disorder-relevant pathways as well as level of protein expression by literature and multiple database queries.

Figure 2.

A, Schematic representation of a comprehensive selection process to identify potential T-cell target antigens. Proteins were considered as identified by fulfilling the criteria for sequence coverage (>10%) and match (n > 5). The engagement in cancer-related disorders and pathways was reviewed by up-to-date literature. Function, localization, and protein expression levels were determined by various online platforms, resulting in a final selection of 79 candidate antigens that were validated further (B) in patients of origin (NCH1150, NCH519a, NCH1390, NCH612). Antigens with a significantly increased immunogenicity (gray bars and asterisks) or 1.3-fold increased levels of immune response (gray bars) were further validated in a bigger study sample of healthy individuals and LGG patients. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

Figure 2.

A, Schematic representation of a comprehensive selection process to identify potential T-cell target antigens. Proteins were considered as identified by fulfilling the criteria for sequence coverage (>10%) and match (n > 5). The engagement in cancer-related disorders and pathways was reviewed by up-to-date literature. Function, localization, and protein expression levels were determined by various online platforms, resulting in a final selection of 79 candidate antigens that were validated further (B) in patients of origin (NCH1150, NCH519a, NCH1390, NCH612). Antigens with a significantly increased immunogenicity (gray bars and asterisks) or 1.3-fold increased levels of immune response (gray bars) were further validated in a bigger study sample of healthy individuals and LGG patients. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

Close modal

Epitope prediction

To be presented as a T-cell epitope, antigens pass various crucial modification processes involved in the MHC-class I antigen presentation pathway. In order to identify the epitopes presented to T cells in MHC-class I–dependent manner, we used the "Proteasomal cleavage/Tap transport/MHC-class I combined predictor" algorithm (http://www.iedb.org). This algorithm generates an overall score characterizing the intrinsic potential of T-cell epitopes to be presented, by predicting the likelihood of peptides to be naturally processed from their protein of origin and to bind to MHC-I molecules. Prediction of MHC binding was manually defined by limiting the HLA alleles to A*01:01, A*02:01, A*24:02, A*03:01, and B*07:02 (23) and selecting the immune proteasomal cleavage type. Peptide sequences (50 amino acids, AA) were selected based on a strong average-estimated binding (IC50) and content of high numbers of possible MHC-binding epitope combinations for the given alleles (IC50 cut-off < 500 nmol/L; Supplementary Table S2). Reactive epitopes with the length of 8, 9, and 10 AA were exclusively predicted for HLA-A*02:01 with an exception for CNTN1. Peptides were synthesized by the Genomic and Proteomics Core Facility (German Cancer Research Center, Heidelberg, Germany).

Cells and culture conditions

In-house–generated IDHmut glioma stem-like cells (GSC) NCH551b, NCH612, NCH620, NCH645, and NCH1681 were maintained in serum-free medium as described (19). Authentication of all cell line stocks was done by short tandem repeat profiling (DMSZ). Cell lines were controlled for the absence of mycoplasma contamination on a regular basis.

Generation of human T cells and DCs

Heparin-buffered human peripheral blood samples were obtained preoperatively by venipuncture and were further processed within 8 hours of collection. Peripheral blood mononuclear cells (PBMC) were prepared using a Ficoll gradient (#L6113; Biochrom) following the manufacturer's instructions, but without dilution of whole blood. For quantification of viability, the cells were stained with Trypan blue dye (#T8154; Sigma-Aldrich) and counted in a Neubauer counting chamber. DCs were generated as described previously with modifications (24). Briefly, adherent PBMCs were cultured for 5 to 7 days in serum-free X-VIVO 20 (#BE04-448Q; Lonza) containing 50 ng/mL rhuGM-CSF (#NDC58468-0181-2; Genzyme) and 1,000 U/mL IL4 (#130-093-924; Miltenyi Biotec). DCs were enriched using negative selection with Dynabeads Pan Mouse IgG (#11041; Invitrogen) beads coupled with anti-CD3, anti-CD19 (both G. Moldenhauer, DKFZ), and anti-CD56 (#IM1844; Beckman Coulter) antibodies. The resulting purified DCs were further used as antigen-presenting cells in the subsequent IFNγ ELISpot assays. Autologous peripheral blood mononuclear T cells (PBMTCs) were isolated from nonadherent PBMCs after 5 to 7 days of cultivation in serum-free X-VIVO 20 containing 100 U/mL IL2 (#02238131; Novartis) and 60 U/mL IL4 in order to prevent natural killer cell expansion without additional restimulation or expansion by using the Dynabeads Untouched Human T-cells Kit (#11344D; Invitrogen).

IFNγ ELISpot

IFNγ secretion of T cells was determined as described (22). In brief, DCs were pulsed with either protein fractions (25 μL) or long peptides (50 AA; 200 μg antigen per 2 × 105 cells/mL) in cytokine-free X-VIVO 20 overnight. When short peptides (reactive epitopes; 8–10 AA) were used, DCs were pulsed for 3 to 4 hours (20 μg antigen per 2 × 105 cells/mL). Subsequently, antigen-pulsed and -unpulsed DCs were incubated with purified autologous T cells (ratio = 1:5) for 40 hours. Evaluation of the IFNγ plates was performed using a CTL Analyzer and SmartCount, an automated counting protocol by the ImmunoSpot-Software (both Cellular Technology Ltd.). All obtained counts were manually audited and reviewed. IFNγ spot numbers observed in the negative control settings [DCs pulsed with lysate from peripheral blood lymphocytes (PBL), peptides derived from IgG1 (50 AA, for long peptides), or HIV (gag/pol; 9 AA, for reactive epitopes)] were considered as a threshold, to quantify antigen-specific T-cell responses. Responses were defined as significantly higher numbers of IFNγ spots in test wells as compared with the respective control wells, if not otherwise specified. Data for all patients of origin are shown in Fig. 2B (Supplementary Fig. S2).

Cytokine capture assay

The cytokine capture assay was performed according to the manufacturer's instructions using the TNFα and IFNγ Secretion Assay Detection Kits (#130-091-267 and #130-090-433; Miltenyi Biotec). In brief, DCs of 3 patients were pulsed with antigens or IgGs (Kiovig; #PZN-4668835; Baxter Deutschland) for 8 to 10 hours, and then cocultivated with preincubated autologous T cells (ratio = 1:5) for 12 hours as described above. Then, cells were collected, and unspecific binding was blocked by incubation with 1.7 mg/mL Kiovig for 10 minutes on ice. Subsequently, PBMTCs were harvested (95%–98% purity) and labeled with TNFα and IFNγ catch reagents supplied by the manufacturer. Cells were incubated for 45 minutes at 37°C with tilting and rotation to minimize unspecific cytokine binding throughout cytokine secretion while supporting the immediate binding of TNFα or IFNγ to the surface of the respective T cells by TNFα- or IFNγ-bispecific antibodies. Secreted cytokines captured on the surface of T cells were detected by fluorochrome-conjugated detection antibodies (IFNγ–FITC, TNFα–APC). Live/dead cells were discriminated by staining with LIVE/DEAD Fixable Yellow Dead Cell Stain (1:1,000; #L34967, Thermo Fisher Scientific) followed by a T-cell subset staining using CD4-PerCP-Cy5.5 (L200; #552838, BD Biosciences), CD8-V450 (RPA-T8; #560347, BD Biosciences), CD62L-APC-Cy7 (DREG-56; #304814, BioLegend), and CD45RA-PE-Cy7 (5H9; #561216, BD Biosciences).

Peptide-MHC-I human tetramers

Quantification of antigen-specific peripheral CD8+ T cells was performed using iTAg Custom Tetramers (HLA-A*02:01; pMHC) supplied by MBL International Corporation. In the following, customized phycoerythrin (PE)-conjugated tetramers (MBL-T02001) loaded with reactive epitopes were used. An appropriate positive control (MBL-T0111; Influenza-M1: GILGFVFTL) as well as negative control (MBL-T01044; Negative Tetramer) was used for validation and to define background signal. PBMTC samples (1 × 106 cells/mL per sample) were distributed at 200 mL/well in Greiner CELLSTAR 96-well plates (#M3562, Sigma-Aldrich), blocked with Kiovig, and subsequently costained with LIVE/DEAD Fixable Yellow Dead Cell Stain Kit, 10 μL customized pMHC tetramers, 5 μL CD3 Alexa Fluor 700 (OKT-3; #317340, BioLegend), and 3 μL CD8 Alexa Fluor 488 (RPA-T8; #301021, BioLegend) antibodies. Cells were incubated for 30 minutes at room temperature. Pellet was resuspended in 500 μL PBS [0.5% PFA (#P6148, Sigma-Aldrich)] and stored at 4°C for 1 to 24 hours prior to analysis. Analysis was performed using a FACS Aria II and the FACS Diva software (both BD Biosciences) in the FACS Core Facility (University Hospital, Heidelberg). The results are visualized as percentage (%) of pMHC tetramer–bound cytotoxic T cells (PEpositiveCD3+CD8+) among PEnegativeCD3+CD8+ T cells. Sorted tetramer+CD3+CD8+ T cells were collected in 1.5 mL LoBind microcentrifuge tubes (#0030108116, Eppendorf), pelleted, and frozen at −80°C for subsequent molecular analyses. All steps were conducted under exploratory research conditions according to previously established protocols.

HLA typing

HLA typing of all patients was performed by the Department of Transplantation Immunology at the University Hospital of Heidelberg (Heidelberg, Germany), as previously described (25).

Immunohistochemical stainings

Acetone-fixed LGG cryosections (5–7 μm) were stained with the optimized concentration of primary antibodies [anti-CRK (#H00001398-D01P; Abnova); anti-CNTN1 (#ABIN571818; antibodies-online); anti-NME2 (#NBP2-19553; Novousbio); anti-CFL1 (#GTX102156; GeneTex)] and the appropriate isotype controls IgG1 (#ab91353; Abcam) and IgG (#ab27478; Abcam) for 1 hour followed by 3 times washing [1x DPBS (#14190094; Invitrogen); 0.05% Tween20 (#P1379; Sigma-Aldrich)]. Detection was carried out by the anti-mouse and anti-rabbit detection system Vectastain Elite ABC Kit (#PK-6102 and #PK-6101; both Vector Laboratories) utilizing all reagents according to the manufacturer's protocol. Immunohistochemistry for TKT was conducted on 4-μm-thick formalin-fixed, paraffin-embedded tissue sections mounted on StarFrost Advanced Adhesive Slides (Engelbrecht) followed by drying at 80°C for 15 minutes. Immunohistochemistry was performed on a Bench-Mark Ultra immunostainer (Ventana Medical Systems). After deparaffinization, slides were pretreated at 95°C in Cell Conditioning 1 Buffer (#950-124; Ventana) for 92 minutes. The sections were incubated with primary anti-TKT antibody (HPA029480; Sigma-Aldrich, diluted 1:2,000) for 2 hours. Standard Ventana signal amplification was used.

Flow cytometry

In order to quantify antigen expression in IDHmut GSCs, 5 × 105 live cells were stained with anti-CRK, anti-CNTN1, anti-NME2, anti-TKT, and anti-CFL1 antibodies as described (see immunohistochemical stainings), which were conjugated to a fluorescent dye by preincubation with respective species-specific PE-conjugated AffiniPureF(ab)2 fragments (#115-116-146 or #111-116-144; both Jackson ImmunoResearch). Appropriate isotype controls (#ab27478 or #ab91353; both Abcam) were used according to the manufacturer's instructions. Cells were analyzed by flow cytometry using a FACS LSR II and the FACS Diva software (both BD Biosciences). Antigen-expressing cells were defined by mean-staining fluorescence intensity, above the level of the respective isotype control.

The Cancer Genome Atlas data

Level 3 RNASeq gene expression data for GBMs (26) and LGGs (2) including the appropriate clinical data were downloaded from The Cancer Genome Atlas (TCGA) website (http://cancergenome.nih.gov). Genes were normalized to root-mean-square deviation and subsequently voom-transformed. Solely primary IDHmut LGGs were used for analysis and further separated in astrocytomas (IDHmut 1p/19q no-codeletion) and oligodendrogliomas (IDHmut and 1p/19q codeletion) and the corresponding WHO grade, as acquired from the clinical data.

Statistical analysis

GraphPad Prism (Ver.6) was used for statistical analyses of ELISpot data. All other analyses were conducted by means of the R software environment (Ver.3.4.1). Statistical evaluation was done using a two-sided Student t test. P values ≤ 0.05 were considered significant. If not indicated otherwise, error bars show the SEM of 3 to 5 replicate wells.

Identification of immunogenic antigens in protein fractions

To identify TAAs potentially triggering endogenous T-cell responses within complex tumor proteomes of IDHmut LGGs (n = 4), we applied proteomics using the PF2D system in combination with IFNγ ELISpot assays (Fig. 1A; Supplementary Table S1; ref. 22). Tumor protein lysates were subjected to the chromatofocusing first-dimension fractionation. Resulting protein fractions were tested for recognition by T cells in a completely autologous IFNγ ELISpot assay, in which DCs were pulsed with 1D fractions or with lysate from autologous PBL as negative control. Figure 1B exemplifies the identification of five fractions recognized by T cells (significant IFNγ spot counts as compared with PBL), which were further separated based on hydrophobicity in the second dimension. Obtained 2D fractions were again tested for T-cell recognition by ELISpot (Fig. 1C). Four of the five immunogenic 1D fractions resulted in one to three immunogenic 2D fractions. Tumor tissues from 3 additional patients were analyzed accordingly (Supplementary Fig. S1A and S1B). To determine the protein composition of immunogenic 2D fractions, LC-ESI-MS/MS analysis was performed revealing altogether 2,897 potential TAA candidates based on the following criteria: detected peptides covering more than 10% of theoretical protein sequence and a minimum of five peptides matching to the corresponding protein (Fig. 2A). After exclusion of frequent contaminations such as keratins (21), remaining candidates were characterized regarding their function using the publicly available databases Uniprot (27) and GeneCards (www.genecards.org). In addition, a comprehensive literature research was performed to narrow down the list of potential TAA candidates based on their assumed role in tumor diseases or tumor-specific signaling pathways. Finally, candidates displaying a strong protein staining in normal brain (NB) tissues according to the Human Protein Atlas were excluded (28). Based on this filtering process, 79 proteins detectable in immunogenic proteome fractions of IDHmut glioma lysates were selected for further testing.

Validation of antigen recognition in patients of origin

To verify whether selected proteins were recognized by endogenous T cells, we predicted in silico for each protein long peptide sequences (50 AA) containing a high number of potential T-cell epitopes estimated by high MHC-I–binding affinity (IC50 cut-off < 500 nmol/L) using NetMHCpan (Supplementary Table S2). Long peptides were synthesized and tested in IFNγ ELISpot assays with T cells from respective patients. By this approach, we were able to confirm the presence of endogenous antigen-specific T cells in 3 of 4 patients tested (numbers of immunogenic antigens: NCH1150: 8/24, NCH1390: 11/24, NCH612: 1/17, respectively) as demonstrated by significantly higher numbers of IFNγ spots in test wells as compared with control wells (Fig. 2B; Supplementary Fig. S2).

Identification of TAAs in LGG

Next, we tested newly identified immunogenic T-cell antigens (n = 20) as well as another six antigens presenting with clearly increased IFNγ spot counts (at least 1.3-fold; Fig. 2B), which might have not reached significance due to variability among triplicates for recognition by endogenous T cells from 16 unrelated LGG patients and 13 healthy individuals (Supplementary Table S3). This analysis revealed significantly higher T-cell responses (IFNγ spot counts in test wells relative to IgG1 control) in LGG patients (Fig. 3C), but no differences between astrocytoma and oligodendroglioma patients (Supplementary Fig. S3A and S3B). Twenty-four of 26 antigens (88.5%) were recognized by preexisting T cells in at least one additional LGG patient, and T cells from 7 patients recognized more than five antigens simultaneously (Fig. 3A and B). Interestingly, five antigens (CRKII, CFL1, CNTN1, NME2, and TKT) induced frequent T-cell responses in 28% to 56% of all patients (Supplementary Fig. S3C) but in no (CRKII, CFL1, and CNTN1) or only one healthy donor (7.7%, NME2 and TKT). These responses were not only significantly higher as compared with healthy individuals (Fig. 3D), but they also reflect the diversity of immune responses among patients as previously has been shown in a similar way in IDHwt GBM (19).

Figure 3.

A, IFNγ ELISpot studies in LGG patients (n = 16) validated the immunogenic potential of CRKII, CFL1, CNTN1, NME2, and TKT, triggering a spontaneous T-cell IFNγ response in more than 25% of all patients tested. B, Although CRKII, CFL1, and CNTN1 triggered no immune response in healthy individuals, 1 of 13 tested patients (7%) featured IFNγ-secreting T cells against TKT and NME2. C, Significant increased total immune response in LGG patients (P) vs. healthy donors (HD) which is in particular seen for the antigens (D) CRKII, CFL1, CNTN1, NME2, and TKT. E, Proportion (%) of IFNγ- and TNFα-secreting CD4+ (top row) and CD8+ (bottom row) peripheral blood mononuclear T cells stimulated with peptides corresponding to potential TAAs, staphylococcal enterotoxin B (SEB), or IgG1 to assign background reactivity. Each symbol represents data from an individual patient (n = 3). *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Figure 3.

A, IFNγ ELISpot studies in LGG patients (n = 16) validated the immunogenic potential of CRKII, CFL1, CNTN1, NME2, and TKT, triggering a spontaneous T-cell IFNγ response in more than 25% of all patients tested. B, Although CRKII, CFL1, and CNTN1 triggered no immune response in healthy individuals, 1 of 13 tested patients (7%) featured IFNγ-secreting T cells against TKT and NME2. C, Significant increased total immune response in LGG patients (P) vs. healthy donors (HD) which is in particular seen for the antigens (D) CRKII, CFL1, CNTN1, NME2, and TKT. E, Proportion (%) of IFNγ- and TNFα-secreting CD4+ (top row) and CD8+ (bottom row) peripheral blood mononuclear T cells stimulated with peptides corresponding to potential TAAs, staphylococcal enterotoxin B (SEB), or IgG1 to assign background reactivity. Each symbol represents data from an individual patient (n = 3). *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Close modal

To characterize antigen-specific T-cell responses against the five most immunogenic peptides more extensively, we performed cytokine secretion assays for IFNγ and TNFα coupled with CD4+/CD8+ T-cell subset characterization (n = 3) which overall revealed antigen-reactive T-cell frequencies comparable with those obtained by IFNγ ELISpot. CFL1 showed a strong (>6-fold over neg. control) IFNγ response in at least 1 patient, whereas NME2 triggered strong responses in all patients (Fig. 3E; Supplementary Fig. S4). All peptides triggered TNFα secretion, yet at varying proportions among patients and peptides that exceeded negative control values by at least 3-fold overall suggesting the occurrence of antigen-reactive mixed CD4 and CD8 T-cell responses against these antigens (mean T-cell frequencies: CD4+: NME2: 0.99%, CFL1: 0.76%, TKT: 0.43%, CNTN1: 0.38%, CRKII: 0.33%, IgG: 0.26%, and CD8+: NME2: 1.00%, CFL1: 0.70%, TKT: 0.52%, CRKII: 0.39%, CNTN1: 0.41%, IgG: 0.25%).

In summary, CRKII, CFL1, CNTN1, NME2, and TKT are frequent targets of endogenous CD4+ and CD8+ T-cell responses in LGG patients but not in healthy individuals.

Expression of immunogenic antigens in tumors and NB tissues

We next studied the expression of CRKII, CFL1, CNTN1, NME2, and TKT by mRNA expression analyses and IHC stainings of IDHmut LGG tissues. To this end, we made use of RNAseq data (LGG TCGA; ref. 2) including 95 IDHmut astrocytomas, 109 IDHmut oligodendrogliomas, and 5 NB tissues (Fig. 4A). We found increased mRNA levels in at least one glioma subgroup for 4 of 5 candidates. Most uniform overexpression was observed for CRK and NME2, whereas it was more restricted to subgroups for CFL1 and TKT. IHC stainings of IDHmut astrocytomas (WHO°II: n = 10; WHO°III: n = 10), oligodendrogliomas (WHO°II: n = 10; WHO°III: n = 10), and NB tissues (n = 2) confirmed increased expression in varying proportions of tumor cells in astrocytomas and oligodendrogliomas, whereas none of the antigens except TKT was detectable in NB (Fig. 4B, Supplementary Figs. S5 and S6A and S6B). Furthermore, we observed a clear localization of CFL1, CRKII, NME2, and CNTN1 in tumor cells as assessed by GFAP costaining, which was almost absent in tumor-infiltrating CD68+ microglia/macrophages as well as CD31+ endothelial cells (Supplementary Fig. S7). In addition, we interrogated a correlation between TAA expression and tumor-infiltrating T lymphocytes and identified a positive correlation of NME2 and CD3E (Supplementary Fig. S8). Thus, we detected a tumor-specific expression of CRKII, NME2, and in part for CFL1 and TKT on mRNA and protein level, whereas CNTN1 overexpression was restricted to the protein level. Moreover, expression of NME2 and CD3E was coregulated, suggesting a role in T-cell–dependent processes.

Figure 4.

A, Usage of RNAseq TCGA data for NB tissues (n = 5) and IDHmut LGG tumors to compare mRNA expression levels in astrocytomas (A, n = 95) and oligodendrogliomas (OD, n = 109) among the WHO grades. B, Representative images of immunohistochemical staining of CFL1, CRKII, NME2, CNTN1, and TKT in NB tissue and astrocytomas (WHO°II: n = 10; WHO°III: n = 10) and OD (WHO°II: n = 10; WHO°III: n = 10). Scale bar, 50 μm; scale bar zoom: 20 μm. *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Figure 4.

A, Usage of RNAseq TCGA data for NB tissues (n = 5) and IDHmut LGG tumors to compare mRNA expression levels in astrocytomas (A, n = 95) and oligodendrogliomas (OD, n = 109) among the WHO grades. B, Representative images of immunohistochemical staining of CFL1, CRKII, NME2, CNTN1, and TKT in NB tissue and astrocytomas (WHO°II: n = 10; WHO°III: n = 10) and OD (WHO°II: n = 10; WHO°III: n = 10). Scale bar, 50 μm; scale bar zoom: 20 μm. *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Close modal

Prediction and validation of reactive epitopes

Because all patients listed in Supplementary Table S4 expressed the HLA allele HLA-A*02:01 (Supplementary Table S5), we applied an in silico approach to predict HLA-A*02:01–restricted T-cell epitopes (8–10 AA) as exemplarily shown for NME2 (Fig. 5A) to all five immunogenic proteins (IC50 < 500 nmol/L; NetMHCpan; Supplementary Table S6) and subsequently assessed the presence of epitope-specific T cells by ELISpot analysis. Due to the lack of predicted HLA-A*02:01 epitopes for CNTN1, we selected for the antigen epitopes with an IC50 < 500 nmol/L for other frequent HLA-alleles (A*11:01, B*07:02). Figure 5B shows a representative ELISpot with synthesized peptides corresponding to predicted reactive epitopes and respective positive (CMV-AdV, SEB) and negative controls (HIV). Highest and most frequent antigen-specific T-cell responses were identified for epitopes predicted for CRKII and NME2 (Fig. 5C; Supplementary Fig. S9A). Less frequently recognized epitopes of TKT and CNTN1 still triggered T-cell responses in at least 1 of 7 patients. No significant T-cell responses were observed for CFL1. Noteworthy, IFNγ spot numbers were comparable between astrocytoma and oligodendroglioma patients (Supplementary Fig. S9B).

Figure 5.

A, HLA-binding affinity prediction (for the alleles: HLA-A*01:01, A*02:01, A*24:02, A*03:01, B*07:02) using the “Proteasomal cleavage/TAP transport/MHC-class I combined predictor” algorithm, selection of 50-mer peptides based on the number of epitopes and a high binding affinity (IC50 < 500 nmol/L). Prediction of reactive epitopes (8-, 9-, and 10-mer peptides) based solely on the HLA-A*02:01 allele. B, IFNγ ELISpot of 1 representative patient (NCH1000). Antigens triggering a significantly increased immune response are indicated as gray bars with an asterisk, highlighting the level of significance. C, IFNγ ELISpots of all HLA-A*02:01 patients tested (n = 7) and significantly responding HLA-A*02:01 patients (in %) relative to negative control (HIV gag/pol 9 AA peptide). CMV-AdV, cytomegalovirus-adenovirus; SEB, staphylococcal enterotoxin B; HIV, human immunodeficiency virus. *, P < 0.05.

Figure 5.

A, HLA-binding affinity prediction (for the alleles: HLA-A*01:01, A*02:01, A*24:02, A*03:01, B*07:02) using the “Proteasomal cleavage/TAP transport/MHC-class I combined predictor” algorithm, selection of 50-mer peptides based on the number of epitopes and a high binding affinity (IC50 < 500 nmol/L). Prediction of reactive epitopes (8-, 9-, and 10-mer peptides) based solely on the HLA-A*02:01 allele. B, IFNγ ELISpot of 1 representative patient (NCH1000). Antigens triggering a significantly increased immune response are indicated as gray bars with an asterisk, highlighting the level of significance. C, IFNγ ELISpots of all HLA-A*02:01 patients tested (n = 7) and significantly responding HLA-A*02:01 patients (in %) relative to negative control (HIV gag/pol 9 AA peptide). CMV-AdV, cytomegalovirus-adenovirus; SEB, staphylococcal enterotoxin B; HIV, human immunodeficiency virus. *, P < 0.05.

Close modal

Altogether, we were able to predict and validate several HLA-restricted epitopes of the herein identified immunogenic TAAs. Particularly in silico prediction for NME2, CRKII, and TKT revealed frequent immunogenic HLA-A*02:01 epitopes and led to the selection of CRKII_9AA_256_264, NME2_9AA_77_85, and TKT_10AA_92_99 for further analysis.

Expression of immunogenic antigens in IDHmut GSCs

Because GSCs are held responsible for therapy resistance and tumor recurrence, thus representing a clinically relevant tumor cell subpopulation (29, 30), we tested TAA expression in patient-derived IDHmut GSCs (n = 5; Supplementary Fig. S10A and S10B). Expression of CFL1 was observed in approximately 2% of GSCs derived from WHO°III cases (NCH612 and NCH1681), and up to 45.4% of sGBM-derived GSCs (Supplementary Table S7). Similar results were found for CRKII, CNTN1, and tendentially for NME2-expressing GSCs, whereas TKT expression showed strong interindividual variability (4.7%–84.8% positive GSCs).

Although to a varying degree, CRKII, CFL1, CNTN1, NME2, and TKT were thus expressed in a substantial proportion of patient-derived IDHmut GSCs with a more pronounced expression for most of them in sGBM GSCs.

Quantification of antigen-specific peripheral T cells

Finally, we examined frequencies of antigen-specific CD8+ T cells in peripheral blood of HLA-A*02:01 IDHmut glioma patients (n = 6) targeting identified HLA-A*02:01 epitopes within the TAAs CRKII, NME2, and TKT (Supplementary Table S8). Flow cytometry–based T-cell quantification using peptide-loaded fluorescent MHC-I tetramers (pMHC) as exemplarily shown for patient NCH1150 (Fig. 6; Supplementary Fig. S10C and S10D) revealed frequencies of antigen-specific T cells up to 2.82% CD3+CD8+ T cells for TKT (mean, 0.69%). In all but one tested glioma patients, we further detected CRKII- (mean, 0.59%) and NME2-specific (mean, 0.19%) CD3+CD8+ T cells. Identified populations of CRKII- and TKT-specific cytotoxic T cells were comparable with those of Influenza M1-reactive cytotoxic T cells (31).

Figure 6.

Quantification of antigen-specific cytotoxic CD3+CD8+ T cells by flow cytometry and TAA-loaded [CRKII (ALALEVGEL), NME2 (MVWEGLNVV), and TKT (FLAEAELLNL)] pMHC I tetramers (PE-conjugated) and the respective negative (neg. tetramer) and positive control (influenza M1). Solid line represents mean background of negative control. *, P < 0.05.

Figure 6.

Quantification of antigen-specific cytotoxic CD3+CD8+ T cells by flow cytometry and TAA-loaded [CRKII (ALALEVGEL), NME2 (MVWEGLNVV), and TKT (FLAEAELLNL)] pMHC I tetramers (PE-conjugated) and the respective negative (neg. tetramer) and positive control (influenza M1). Solid line represents mean background of negative control. *, P < 0.05.

Close modal

In summary, peptide-loaded pMHC-I tetramer stainings supported the identification of HLA-A*02–restricted T-cell epitopes of novel TAAs by detecting antigen-specific CD8+ T-cell populations for CRKII, NME2, and TKT in IDHmut glioma patients.

Using the patient's immune system to control tumor growth by specifically supporting preexisting antigen-specific T-cell responses is currently explored in several cancer immunotherapies (32). Although gliomas are regarded as immunologically cold tumors due to a low mutation rate and a low intratumoral T-cell infiltration (33, 34), even in these tumors, immunotherapeutic approaches are gaining increasing attention including cancer vaccines eliciting specific immune responses against TSA- or TAA-bearing tumor cells (13, 14). However, well-characterized T-cell target antigens of potential therapeutic relevance are still very limited and mainly originate from the analysis of IDHwt GBMs (10). Yet, IDHmut LGGs differ substantially from IDHwt GBMs regarding biology, metabolism, and prognosis and thus require a more thorough analysis in search for T-cell target antigens. In the present study, we successfully identified CRKII, CFL1, CNTN1, NME2, and TKT as novel and frequent immunogenic TAAs in IDHmut gliomas, utilizing an unbiased proteomic and immunologic approach. All five antigens triggered the activation of endogenous T cells in a substantial proportion of up to 56% of astrocytic and/or oligodendroglial IDHmut LGG patients, but rarely in healthy donors. Thus, our findings point to a more frequent immunogenicity of these novel TAAs as compared with already known T-cell targets such as IDH1R132H, which only triggers spontaneous CD4+ T-cell responses in about 16% of IDHmut LGG patients (35). Immunogenicity of the antigens was further confirmed in a cytokine capture assay, where we found a rare but consistent population of antigen-reactive T cells. Verification of these patient-specific antitumor T-cell responses is in good accordance with recent perceptions of the central nervous system (CNS) as immune-active organ (36). Indeed, for most of the antigens, robust overexpression was found in gliomas and even in IDHmut GSCs. In contrast to T-cell responses specific for IDH mutations, we here characterized T-cell responses against wild-type self-antigens. These are subject to antigen presentation by thymic epithelial cells and thus to central tolerance mechanisms which might result in reduced T-cell receptor avidity compared with mutation-specific T cells. We thus cannot exclude the possibility that this correlates with reduced functional capacity of T cells reactive to these antigens.

This is one of the first attempts to systemically characterize the repertoire of naturally presented antigens resulting in spontaneous T-cell responses in IDHmut glioma patients. Because treatment of LGGs still remains unsatisfactory, novel T-cell targets are of particular interest to develop successful immunotherapeutics (37). Consistent with the NCI's prioritization, we based the selection of novel T-cell targets on the endogenous immunogenicity of potential TAAs (18).

Beyond immunogenicity, all novel T-cell targets correspond to further characteristics as defined by the NCI (18). That is, all identified proteins showed a specific protein expression in IDHmut lower-grade astrocytic and oligodendroglial tumors and also on IDHmut GSCs which appear to be indispensable for tumor recurrence (29, 30). Verified expression of all TAAs on both tumor-initiating cells and tumor bulk cells is an essential requirement for cancer vaccine target antigens by increasing the chance for a more effective treatment and by simultaneously reducing the risk of an immune escape through targeting an extended spectrum of tumor cells. Accordingly, clinical results of a trial targeting solely one antigen on GBM cells (EGFRvIII) by peptide vaccination demonstrated specific eradication of the intended antigen-expressing tumors cells but failed to improve survival presumably due to the ongoing growth of antigen-negative cells (38). Thus, the observed intratumoral heterogeneity as well as interpatient variances of antigen expression in IDHmut gliomas endorses the usage of a multipeptide vaccine using several T-cell target antigens to overcome the limitation of tumor heterogeneity and tumor escape–related antigen loss.

Noteworthy, all promising TAAs are involved in cancer-relevant pathways. The oncogene CRKII functions in various phosphotyrosine kinase signaling pathways and plays an important role in the regulation of cell adhesion, spreading, and migration through interaction with Rho/Rac proteins (39). Consistent with the observed overexpression in IDHmut gliomas and GSCs, CRKII is highly expressed in various human cancers including GBM (40, 41). The actin-binding protein CFL1 is implicated in invasion, migration, and radiosensitivity of GBMs, metastasis formation in mammary tumors, and apoptotic processes in prostate cancer by contributing to cytoskeletal organization (42–44). Likewise, its overexpression serves as biomarker in non–small cell lung cancer (45). CNTN1 is involved in cell–matrix interaction, and in line with our staining data, its overexpression was associated with glioma malignancy (46, 47). Although NME2 functions as metastasis suppressor in various cancers by regulating tumor cell proliferation, it has been described as a TAA in chronic myeloid leukemia (48–50).

To our knowledge, no other study so far investigated the immunogenic potential of these TAAs in the context of IDHmut LGGs or even of non-CNS tumors. The additional characterization of HLA-A*02–restricted T-cell epitopes and the detection of a considerable percentage of antigen-reactive peripheral T cells empower their usage as targets for immunotherapy in a high proportion of LGG patients. Reactivation of the preexisting repertoire of antigen-specific T cells following the standard surgical resection to minimize the immunosuppressive tumor microenvironment might help to overcome the lack of substantial numbers of neoepitopes and thus render these immunologically cold tumors immune-sensitive (34). Besides, the identified sequences of the reactive epitopes for CRKII, NME2, and TKT could serve to monitor endogenous and treatment-induced glioma-specific T-cell responses in an immunotherapeutic setting independent of the therapeutic use of the newly identified TAAs.

However, this study is limited as it focuses on the identification of novel immunogenic antigens and related CD4+ and CD8+ T-cell responses in a broad proportion of IDHmut gliomas, whereas it does not provide in vivo proof of concept for the therapeutic relevance of active immunization against the uncovered antigens. Nevertheless, by assessing the entire proteome of IDHmut LGGs, while looking for naturally immunogenic LGG-associated antigens, we were able to identify and validate five novel T-cell target antigens with a remarkable tumor-selective immunogenicity. All antigens were expressed in WHO°II and WHO°III astrocytic and oligodendroglial tumors as well as on IDHmut GSCs, and HLA-A*02–restricted T-cell epitopes for CRKII, NME2, and TKT could be conclusively identified. Despite its limitations, our study thus considerably extends the necessary repertoire of potential target antigens for immunotherapy of IDH-mutant glioma.

A. von Deimling reports receiving commercial research support from Bayer-Schering. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Dettling, S. Stamova, P. Beckhove, C. Herold-Mende

Development of methodology: S. Dettling, S. Stamova, C. Rapp, N. Grabe, A. von Deimling, C. Herold-Mende

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Dettling, S. Stamova, M. Schnölzer, A. Rathinasamy, D. Reuss, S. Roesch, C. Jungk, U. Warnken, C. Schramm, A. von Deimling, C. Herold-Mende

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Dettling, S. Stamova, R. Warta, M. Schnölzer, C. Rapp, A. Rathinasamy, K. Pocha, U. Warnken, M.A. Weigand, A. von Deimling, C. Herold-Mende

Writing, review, and/or revision of the manuscript: S. Dettling, S. Stamova, R. Warta, C. Rapp, S. Roesch, C. Jungk, M.A. Weigand, A. Unterberg, P. Beckhove, C. Herold-Mende

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V. Eckstein, A. von Deimling, A. Unterberg, C. Herold-Mende

Study supervision: R. Warta, M.A. Weigand, P. Beckhove, C. Herold-Mende

This work was supported by the Anni-Hofmann-Stiftung. The authors thank the Tissue Bank of the National Center for Tumor Diseases (NCT, Heidelberg, Germany) for providing tissue samples. They also thank M. Greibich, F. Kashfi, I. Hearn, H. Goeltzer, A. Schoeffel, C. Sackmann (Division of Experimental Neurosurgery), S. Juenger and M. Bucur (Translational Immunology Department), and R. Mayer (Functional Proteome Analysis) for excellent technical assistance.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Louis
DN
,
Perry
A
,
Reifenberger
G
,
Deimling von
A
,
Figarella-Branger
D
,
Cavenee
WK
, et al
The 2016 World Health Organization classification of tumors of the central nervous system: a summary
.
Acta Neuropathol
2016
;
131
:
803
20
.
2.
Brat
DJ
,
Verhaak
RGW
,
Al-dape
KD
,
Yung
WKA
,
Salama
SR
,
Cooper
LAD
, et al
Comprehensive, integrative genomic analysis of diffuse lower-grade gliomas
.
N Engl J Med
2015
;
372
:
2481
98
.
3.
Dang
L
,
White
DW
,
Gross
S
,
Bennett
BD
,
Bittinger
MA
,
Driggers
EM
, et al
Cancer-associated IDH1 mutations produce 2-hydroxyglutarate
.
Nature
2009
;
462
:
739
44
.
4.
Turcan
S
,
Rohle
D
,
Goenka
A
,
Walsh
LA
,
Fang
F
. 
IDH1 mutation is sufficient to establish the glioma hypermethylator phenotype
.
Nature
2012
;
483
:
479
83
.
5.
Wen
PY
,
Reardon
DA
. 
Neuro-oncology in 2015: Progress in glioma diagnosis, classification and treatment
.
Nature Rev Neurol
2016
;
12
:
69
70
.
6.
van den Bent
MJ
. 
Practice changing mature results of RTOG study 9802: another positive PCV trial makes adjuvant chemotherapy part of standard of care in low-grade glioma
.
Neuro Oncol
2014
;
16
:
1570
4
.
7.
Postow
MA
,
Chesney
J
,
Pavlick
AC
,
Robert
C
,
Grossmann
K
,
McDermott
D
, et al
Nivolumab and ipilimumab versus ipilimumab in untreated melanoma
.
N Engl J Med
2015
;
372
:
2006
17
.
8.
Carson
MJ
,
Doose
JM
,
Melchior
B
,
Schmid
CD
,
Ploix
CC
. 
CNS immune privilege: hiding in plain sight
.
Immunol Rev
2006
;
213
:
48
65
.
9.
Plautz
GE
,
Miller
DW
,
Barnett
GH
,
Stevens
GH
,
Maffett
S
,
Kim
J
, et al
T cell adoptive immunotherapy of newly diagnosed gliomas
.
Clin Cancer Res
2000
;
6
:
2209
18
.
10.
Reardon
DA
,
Freeman
G
,
Wu
C
,
Chiocca
EA
,
Wucherpfennig
KW
,
Wen
PY
, et al
Immunotherapy advances for glioblastoma
.
Neuro Oncol
2014
;
16
:
1441
58
.
11.
Lawrence
MS
,
Stojanov
P
,
Polak
P
,
Kryukov
GV
,
Cibulskis
K
,
Sivachenko
A
, et al
Mutational heterogeneity in cancer and the search for new cancer-associated genes
.
Nature
2013
;
499
:
214
8
.
12.
Zhang
JG
,
Eguchi
J
,
Kruse
CA
,
Gomez
GG
,
Fakhrai
H
,
Schroter
S
, et al
Antigenic profiling of glioma cells to generate allogeneic vaccines or dendritic cell-based therapeutics
.
Clin Cancer Res
2007
;
13
:
566
75
.
13.
Okada
H
,
Kalinski
P
,
Ueda
R
,
Hoji
A
,
Kohanbash
G
,
Donegan
TE
, et al
Induction of CD8+ T-cell responses against novel glioma-associated antigen peptides and clinical activity by vaccinations with {alpha}-type 1 polarized dendritic cells and polyinosinic-polycytidylic acid stabilized by lysine and carboxymethylcellulose in patients with recurrent malignant glioma
.
J Clin Oncol
2011
;
29
:
330
6
.
14.
Okada
H
,
Butterfield
LH
,
Hamilton
RL
,
Hoji
A
,
Sakaki
M
,
Ahn
BJ
, et al
Induction of robust type-I CD8+ T-cell responses in WHO grade 2 low-grade glioma patients receiving peptide-based vaccines in combination with poly-ICLC
.
Clin Cancer Res
2015
;
21
:
286
94
.
15.
Ge
L
,
Cornforth
AN
,
Hoa
NT
,
Delgado
C
,
Chiou
SK
,
Zhou
YH
, et al
Differential glioma-associated tumor antigen expression profiles of human glioma cells grown in hypoxia.
PLoS One
2012
;
7
:
e42661
.
16.
Liau
LM
,
Lallone
RL
,
Seitz
RS
,
Buznikov
A
,
Gregg
JP
. 
Identification of a human glioma-associated growth factor gene, granulin, using differential immuno-absorption
.
Cancer Res
2000
;
60
:
1353
60
.
17.
Dutoit
V
,
Herold-Mende
C
,
Hilf
N
,
Schoor
O
,
Beckhove
P
,
Bucher
J
, et al
Exploiting the glioblastoma peptidome to discover novel tumour-associated antigens for immunotherapy
.
Brain
2012
;
135
:
1042
54
.
18.
Cheever
MA
,
Allison
JP
,
Ferris
AS
,
Finn
OJ
,
Hastings
BM
,
Hecht
TT
, et al
The prioritization of cancer antigens: a national cancer institute pilot project for the acceleration of translational research
.
Clin Cancer Res
2009
;
15
:
5323
37
.
19.
Rapp
C
,
Warta
R
,
Stamova
S
,
Nowrouzi
A
,
Geisenberger
C
,
Gal
Z
, et al
Identification of T cell target antigens in glioblastoma stem-like cells using an integrated proteomics-based approach in patient specimens
.
Acta Neuropathol
2017
;
132
:
1
20
.
20.
Aretz
S
,
Krohne
TU
,
Kammerer
K
,
Warnken
U
,
Hotz-Wagenblatt
A
,
Bergmann
M
, et al
In-depth mass spectrometric mapping of the human vitreous proteome
.
Proteome Sci
2013
;
11
:
22
.
21.
Hodge
K
,
Have
ST
,
Hutton
L
,
Lamond
AI
. 
Cleaning up the masses: exclusion lists to reduce contamination with HPLC-MS/MS
.
J Proteomics
2013
;
88
:
92
103
.
22.
Beckhove
P
,
Warta
R
,
Lemke
B
,
Stoycheva
D
,
Momburg
F
,
Schnölzer
M
, et al
Rapid T cell-based identification of human tumor tissue antigens by automated two-dimensional protein fractionation
.
J Clin Invest
2010
;
120
:
2230
42
.
23.
Gonzalez-Galarza
FF
,
Takeshita
LYC
,
Santos
EJM
,
Kempson
F
,
Maia
MHT
,
da Silva
ALS
, et al
Allele frequency net 2015 update: new features for HLA epitopes, KIR and disease and HLA adverse drug reaction associations
.
Nucleic Acids Res
2015
;
43
:
D784
8
.
24.
Feuerer
M
,
Beckhove
P
,
Bai
L
,
Solomayer
E-F
,
Bastert
G
,
Diel
IJ
, et al
Therapy of human tumors in NOD/SCID mice with patient-derived reactivated memory T cells from bone marrow
.
Nat Med
2001
;
7
:
452
8
.
25.
Heinold
A
,
Bauer
M
,
Scherer
S
,
Opelz
G
,
Tran
TH
. 
Characterization of a new HLA-B allele, HLA-B*5312, and re-evaluation of the published sequences of the untranslated regions of HLA-B*35 and HLA-B*53
.
Tissue Antigens
2007
;
70
:
319
23
.
26.
Brennan
CW
,
Verhaak
RGW
,
McKenna
A
,
Campos
B
,
Noushmehr
H
,
Salama
SR
, et al
The somatic genomic landscape of glioblastoma
.
Cell
2013
;
155
:
462
77
.
27.
UniProt Consortium
. 
UniProt: a hub for protein information
.
Nucleic Acids Res
2015
;
43
:
D204
12
.
28.
Uhlen
M
,
Fagerberg
L
,
Hallström
BM
,
Lindskog
C
,
Oksvold
P
,
Mardinoglu
A
, et al
Tissue-based map of the human proteome
.
Science
2015
;
347
:
1260419
9
.
29.
Singh
SK
,
Hawkins
C
,
Clarke
ID
,
Squire
JA
,
Bayani
J
,
Hide
T
, et al
Identification of human brain tumour initiating cells
.
Nature
2004
;
432
:
396
401
.
30.
Chen
J
,
McKay
RM
,
Parada
LF
. 
Malignant glioma: lessons from genomics, mouse models, and stem cells
.
Cell
2012
;
149
:
36
47
.
31.
Frleta
D
,
Yu
CI
,
Klechevsky
E
,
Flamar
A-L
,
Zurawski
G
,
Banchereau
J
, et al
Influenza virus and poly(I:C) inhibit MHC class I-restricted presentation of cell-associated antigens derived from infected dead cells captured by human dendritic cells
.
J Immunol
2009
;
182
:
2766
76
.
32.
Mellman
I
,
Coukos
G
,
Dranoff
G
. 
Cancer immunotherapy comes of age
.
Nature
2012
;
480
:
480
9
.
33.
Lohr
J
,
Ratliff
T
,
Huppertz
A
,
Ge
Y
,
Dictus
C
,
Ahmadi
R
, et al
Effector T-cell infiltration positively impacts survival of glioblastoma patients and is impaired by tumor-derived TGF
.
Clin Cancer Res
2011
;
17
:
4296
308
.
34.
Schumacher
TN
,
Schreiber
RD
. 
Neoantigens in cancer immunotherapy
.
Science
2015
;
348
:
69
74
.
35.
Schumacher
T
,
Bunse
L
,
Pusch
S
,
Sahm
F
,
Wiestler
B
,
Quandt
J
, et al
A vaccine targeting mutant IDH1 induces antitumour immunity
.
Nature
2014
;
512
:
324
7
.
36.
Heimberger
AB
,
Sampson
JH
. 
Immunotherapy coming of age: what will it take to make it standard of care for glioblastoma?
Neuro Oncol
2011
;
13
:
3
13
.
37.
Sloan
AE
,
Okada
H
,
Ryken
TC
,
Kalkanis
SN
,
Olson
JJ
. 
The role of emerging therapy in the management of patients with diffuse low grade glioma
.
J Neurooncol
2015
;
125
:
631
5
.
38.
Tivnan
A
,
Heilinger
T
,
Lavelle
EC
,
Prehn
JHM
. 
Advances in immunotherapy for the treatment of glioblastoma
.
J Neurooncol
2017
;
131
:
1
9
.
39.
Gu
J
,
Sumida
Y
,
Sanzen
N
,
Sekiguchi
K
. 
Laminin-10/11 and fibronectin differentially regulate integrin-dependent Rho and Rac activation via p130(Cas)-CrkII-DOCK180 pathway
.
J Biol Chem
2001
;
276
:
27090
7
.
40.
Nishihara
H
,
Tanaka
S
,
Tsuda
M
,
Oikawa
S
,
Maeda
M
. 
Molecular and immunohistochemical analysis of signaling adaptor protein Crk in human cancers
.
Cancer Lett
2002
;
180
:
55
61
.
41.
Wang
L
,
Tabu
K
,
Kimura
T
,
Tsuda
M
,
Linghu
H
,
Tanino
M
, et al
Signaling adaptor protein Crk is indispensable for malignant feature of glioblastoma cell line KMG4
.
Biochem Biophys Res Commun
2007
;
362
:
976
81
.
42.
Shaheed
S-U
,
Rustogi
N
,
Scally
A
,
Wilson
J
,
Thygesen
H
,
Loizidou
MA
, et al
Identification of stage-specific breast markers using quantitative proteomics
.
J Proteome Res
2013
;
12
:
5696
708
.
43.
Du
HQ
,
Du
HQ
,
Chen
L
,
Chen
L
,
Wang
Y
,
Wang
Y
, et al
Increasing radiosensitivity with the downregulation of cofilin-1 in U251 human glioma cells
.
Mol Med Rep
2015
;
11
:
3354
60
.
44.
Lu
LI
,
Fu
NI
,
Luo
XU
,
Li
XY
,
Li
XP
. 
Overexpression of cofilin 1 in prostate cancer and the corresponding clinical implications
.
Oncol Lett
2015
;
9
:
2757
61
.
45.
Castro
MAA
,
Dal Pizzol
F
,
Zdanov
S
,
Soares
M
,
Müller
CB
,
Lopes
FM
, et al
CFL1 expression levels as a prognostic and drug resistance marker in nonsmall cell lung cancer
.
Cancer
2010
;
116
:
3645
55
.
46.
Falk
J
,
Bonnon
C
,
Girault
J-A
,
Faivre-Sarrailh
C
. 
F3/contactin, a neuronal cell adhesion molecule implicated in axogenesis and myelination
.
Biol Cell
2002
;
94
:
327
34
.
47.
Eckerich
C
,
Zapf
S
,
Ulbricht
U
,
Müller
S
,
Fillbrandt
R
,
Westphal
M
, et al
Contactin is expressed in human astrocytic gliomas and mediates repulsive effects
.
Glia
2006
;
53
:
1
12
.
48.
Hennessy
C
,
Henry
JA
,
May
FEB
,
Westley
BR
,
Angus
B
,
Lennard
TWJ
. 
Expression of the antimetastatic gene nm23 in human breast cancer: an association with good prognosis
.
J Natl Cancer Inst
1991
;
83
:
281
5
.
49.
Jung
S
,
Paek
Y-W
,
Moon
K-S
,
Wee
S-C
,
Ryu
H-H
,
Jeong
Y-I
, et al
Expression of Nm23 in gliomas and its effect on migration and invasion in vitro
.
Anticancer Res
2006
;
26
:
249
58
.
50.
Tschiedel
S
,
Bach
E
,
Jilo
A
,
Wang
S-Y
,
Lange
T
,
Al-Ali
H-K
, et al
Bcr–Abl dependent post-transcriptional activation of NME2 expression is a specific and common feature of chronic myeloid leukemia
.
Leuk Lymphoma
2012
;
53
:
1569
76
.

Supplementary data