Main

To explore the immunogenicity of IDH1(R132H), 10-mer and 15-mer peptide libraries were generated encompassing the mutated (R132H) and the corresponding wild-type residue of human IDH1 (Extended Data Fig. 1a–c). Neither interrogation of MHC binding prediction algorithms nor T2 binding assays (Extended Data Fig. 1d–f) demonstrated peptide binding to the common MHC class I allele human leukocyte antigen (HLA)-A*0201. In contrast, as shown by MHC class II binding assay, 15-mer peptides bound to the common MHC class II allele HLA-DRB1*0101 (Fig. 1a and Extended Data Fig. 1g). Vaccination of MHC-humanized HLA-A*0201 HLA-DRA*0101 HLA-DRB1*0101 transgenic mice devoid of mouse MHC (A2.DR1 mice) (ref. 16) with a p123–142 (R132H) peptide resulted in a robust interferon (IFN)-γ T-cell response corresponding to the demonstrated peptide HLA-DRB1*0101 binding strengths. Importantly, the T-cell response discriminated between mutated and wild-type IDH1 peptides (Fig. 1b, c and Extended Data Fig. 1h, i). T-cell responses were restricted to long peptides, abrogated by an antibody against heterodimeric HLA-DRA (HLA-DR blocking antibody), and not induced by stimulation with 10-mer peptides, suggesting class II-restricted CD4+ T-cell responses (Fig. 1b, c and Extended Data Fig. 1j) with a predominant TH1 phenotype as evidenced by an IDH1(R132H)-specific T-cell line (Fig. 1d–f, Extended Data Figs 2a–c and 3a–c, and Supplementary Note 1). CD4+ T-cell responses are usually associated with an antibody response. Accordingly, in the serum of mice immunized with p123–142 (R132H) but not of control mice, mutation-specific anti-IDH1 (p123-142) antibodies were detectable (Fig. 1g and Extended Data Fig. 4a, b). Collectively, these data indicate that peptides covering the mutated region of human IDH1 induce a mutation-specific CD4+ T-cell and antibody response in a human MHC context.

Figure 1: IDH1(R132H) peptide vaccination induces TH1 and humoral responses in A2.DR1 mice.
figure 1

a, DR1-binding IDH1(R123H) 15-mer epitopes identified by Reveal Class II binding assay. Immediate binding (yellow) and 24-h stability (red) are shown. b, c, ELISpots of IFN-γ splenocyte responses to 15-mers (b) and 10-mers (c) after vaccination with IDH1(R132H) in CFA (red, IDH1(R132H) (p123–142); blue, wild-type IDH1 (p123–142); black, IDH1(R132H) library; p123–142 (R132H versus wild type), Welch t-test; library, ANOVA; n = 3 biological replicates). Error bars, mean + s.e.m. d, e, ELISpots (white, MOG; black, vehicle; red, IDH1(R132H); blue, wild-type IDH1; HLA-A, IDH1(R132H) plus HLA-A blocking antibody; HLA-DR, IDH1(R132H) plus HLA-DR blocking antibody; HLA-A + HLA-DR, IDH1(R132H) plus HLA-A and HLA-DR blocking antibodies; d; Welch t-test; R132H versus wild type; 0.1 µg ml−1 wild type versus MOG, P = 0.41; 1.0 µg ml−1 wild type versus MOG, P = 0.20; n = 3; error bars, mean ± s.e.m.; e; pairwise Welch t-test with Bonferroni correction; n = 2; error bars, mean + s.e.m.). f, IFN-γ secretion assay (flow through, IFN-γ-negative fraction) of an IDH1(R132H)-specific T-cell line. Red numbers in f indicate populations in per cent. n values in d and e indicate number of technical replicates. g, Detection of IDH1(R132H)-specific IgG in serum after vaccination with IDH1(R132H) (red) and vehicle (black) using Montanide (Welch t-test; n = 3 biological replicates). Scatter plot showing individual values. OD, optical density, absorbance.

PowerPoint slide

Full size image

To assess clinical relevance, patients with gliomas and healthy controls (Extended Data Tables 1 and 2) were screened for spontaneous peripheral T-cell responses against IDH1(R132H). In 4 of 25 patients harbouring IDH1(R132H)-mutated gliomas but in none of 29 patients with an IDH1 wild-type glioma, IDH1(R132H)-specific IFN-γ-producing T cells were detectable (Fig. 2a–c). As expected, the IDH1(R132H)-specific T-cell responses detectable in patients with IDH1(R132H)-mutated gliomas were not evoked by the 10-mer peptides (Fig. 2d and Extended Data Fig. 1a, b), were abrogated by an HLA-DR blocking antibody (Fig. 2e) and were predominantly CD4+ (Fig. 2f). HLA typing of IDH1(R132H)-positive glioma patients revealed no clear restriction to a particular HLA type (DRB1*1001;15, DRB1*07;15, DRB1*01;04 and DRB1*03;13, respectively) (Supplementary Note 2 and Extended Data Table 1), supporting the notion that the class II epitope of IDH1(R132H) is promiscuous with respect to HLA class II type as suggested by the vaccination studies in HLA-DRB1*0401 transgenic mice (DR4 mice) (Extended Data Fig. 2a–c). IDH1(R132H)-specific antibodies with predomination of IgG1 subclass were detected in 4 of 42 patients with IDH1(R132H) but not in 47 patients with IDH1 wild-type tumours and not in healthy donors (Fig. 2g, Extended Data Fig. 4c–e and Supplementary Note 2). Collectively, these data indicate that R132H-mutated IDH1 is naturally processed in glioma patients to present an immunodominant epitope in the p123–142 region on MHC class II molecules to CD4+ T cells to induce a spontaneous mutation-specific TH1-polarized response and the production of mutation-specific antibodies detectable in patients with IDH1(R132H)-mutated but not IDH1 wild-type gliomas.

Figure 2: Spontaneous TH1 and humoral IDH1(R132H)-specific responses are detected in IDH1(R132H)+ glioma patients.
figure 2

a, b, Representative ELISpots of IFN-γ PBMC responses of patients p037 (a) and p001 (b) to IDH1 p123–142 (R132H and wild type (WT)). Error bars, mean + s.e.m., n = 2 technical replicates. c, d, ELISpots of IFN-γ PBMC responses of patients with IDH1(R132H)-mutated gliomas (c, red, n = 25; d, n = 23) or wild-type IDH1 gliomas (c, blue, n = 29) to IDH1 p123–142 (c, R132H and wild type) or R132H 10-mer peptides (d). Fisher’s exact test (c). Welch ANOVA; PMA+I, phorbol myristate acetate + ionomycin, positive control (d). Scatter plots showing individual values (after subtraction of MOG-induced spots, negative values set to zero). Dashed line in c indicates cut-off for positivity. e, ELISpot of IFN-γ PBMC responses of patient p037 to IDH1 p123–142 (red, R132H; blue, wild type) including HLA-DR block (Welch t-test; n = 3 technical replicates). Error bars, mean + s.e.m. f, IFN-γ secretion assay of PBMCs from patient p037 (flow through, IFN-γ-negative fraction). Red numbers indicate populations in per cent. g, Detection of IDH1(R132H)-specific IgG in serum of patients with IDH1(R132H) (n = 42), or wild-type IDH1 (n = 47) gliomas, or healthy donors (n = 4) (Fisher’s exact test). Scatter plot showing individual values. Dashed line indicates cut-off for positivity set to 5.

PowerPoint slide

Full size image

Next we analysed processing of the IDH1(R132H)-containing epitope and antitumour effects of the R132H-specific immune response. Whole-tumour cell vaccination of A2.DR1 mice with IDH1(R132H), but not IDH1 wild-type, A2.DR1 sarcomas (Extended Data Fig. 5a–e) induced a robust IDH1(R132H)-specific T-cell response, indicating that IDH1 is indeed endogenously processed to present the R132H-containing epitope on HLA-DRA*0101 HLA-DRB1*0101 (DR1) (Fig. 3a). An inducible antitumour immune response in vivo was demonstrated by vaccination of A2.DR1 mice with IDH1(R132H) (p123–142), which resulted in a growth suppression of subsequently transplanted IDH1(R132H), but not IDH1 wild-type, A2.DR1 sarcomas (Fig. 3b and Extended Data Fig. 5f, g). Importantly, IDH1(R132H)+ tumours, which resisted vaccination, displayed greatly reduced IDH1(R132H) expression whereas sarcomas in the control group retained their IDH1(R132H) expression (Fig. 3c, d). The IDH1(R132H) peptide vaccine also suppressed the growth of pre-established IDH1(R132H)+ sarcomas (Fig. 3e and Extended Data Fig. 5i, j) without any overt toxicity or impairment of IDH1 wild-type enzymatic activity (Extended Data Table 3 and Extended Data Fig. 5h). This therapeutic effect compared well with another HLA-DRB1*0101-restricted peptide vaccine targeting a different, well-studied tumour-associated antigen: NY-ESO-1 (Fig. 3f, Extended Data Fig. 6a–d and Supplementary Note 3).

Figure 3: IDH1(R132H) peptide vaccination reduces IDH1(R132H)+ tumour growth in A2.DR1 mice.
figure 3

a, ELISpot of IFN-γ splenocyte responses to IDH1 p123–142 (R132H (RH) or wild type (WT)) after whole-tumour cell (red, IDH1(R132H)+; black, wild-type IDH1+) vaccination (Welch t-test; n = 4). Scatter plot showing individual values (after subtraction of MOG-induced spots; negative values set to zero) and the mean. b, Growth of subcutaneous sarcomas (IDH1(R132H)+ or wild-type IDH1+) after preventive vaccination with IDH1(R132H) (red) or vehicle (black) (Wilcoxon rank-sum test for median area under the curve (AUC); n = 8 IDH1(R123H), vaccination; n = 6 wild-type IDH1, vaccination; n = 5 IDH1(R132H) and wild type, sham). Error bars, mean ± s.e.m. c, d, IDH1(R132H) expression in subcutaneous sarcomas (IDH1(R132H)+ or wild-type IDH1+) after preventive vaccination with IDH1(R132H) (vaccination, red) or vehicle (sham, black). Representative images (c) and scatter plot showing individual values and the mean (d, Welch one-way ANOVA and pairwise Welch t-tests with Bonferroni correction after log2 transformation; n as in b) are shown. e, f, Growth of pre-established subcutaneous tumours (e, IDH1(R132H)+; f, NY-ESO-1+) after vaccination with IDH1(R132H) (e) or NY-ESO-1 (f) (vaccination, red) or vehicle (black) (Wilcoxon rank-sum test for median AUC; n = 7). Error bars, mean ± s.e.m. All n values indicate numbers of biological replicates.

PowerPoint slide

Full size image

Although there is increasing evidence that antigen-specific cytotoxic CD4+ T cells are in principle capable and sufficient to exert antitumour immunity17,18, the relevance of a CD4+ T-cell-mediated antitumour immune response in the absence of an MHC-class-I-restricted CD8+ T-cell response is still controversial. Analyses of T cells recovered from IFN-γ secretion assays after immunization of A2.DR1 mice with p123–142 (R132H) confirmed that CD4+ but not CD8+ T cells were antigen-specific (Fig. 4a and Extended Data Fig. 7a). In addition, IDH1(R132H)+ A2.DR1 sarcomas were specifically infiltrated by IDH1(R132H) HLA-DRB1*0101 tetramer-positive T cells (Fig. 4b–d). Finally, depletion of CD4+ T cells abrogated the therapeutic effects of the IDH1(R132H) vaccine, indicating that the therapeutic efficacy is dependent on CD4+ T cells (Fig. 4e and Extended Data Fig. 7b). Interestingly, the vaccine-mediated control of tumour growth was also abrogated when CD19+ B cells were depleted, suggesting a contribution of B cells to the therapeutic efficacy of the IDH1(R132H) vaccine (Extended Data Fig. 7c–e and Supplementary Note 4).

Figure 4: Growth suppression of pre-established IDH1(R132H)+ tumours by IDH1(R132H) peptide vaccination is dependent on TH cells.
figure 4

a, IFN-γ secretion assay of splenocytes pooled from 5 A2.DR1 mice vaccinated with IDH1(R132H) or vehicle (sham). b, c, CD3 expression in IDH1(R132H)+ sarcomas after vaccination with IDH1(R132H) (red) or vehicle (sham, black). Representative images (b) and scatter plot showing individual values and the mean (c, Welch one-way ANOVA and pairwise Welch t-tests; n = 5). d, CD4+ T cells in five pooled IDH1(R132H)+ sarcomas (TIL) after vaccination with IDH1(R132H) or vehicle (sham) and IDH1(R132H)-specific T-cell line stained with IDH1(R132H) and control (CLIP) tetramers. e, Growth of subcutaneous IDH1(R132H)+ sarcomas after depletion of CD4+ T cells and vaccination with IDH1(R132H) (blue; depletion + vaccination; red, vaccination; black, sham; Wilcoxon rank-sum test for median AUC with Bonferroni correction; n = 6; n = 7 for vaccination). Error bars, mean ± s.e.m. All n values indicate number of biological replicates.

PowerPoint slide

Full size image

In summary, we show that IDH1(R132H) represents a tumour-specific neoantigen recognized by CD4+ IFN-γ-producing T cells in patients with IDH1(R132H) tumours and that an IDH1(R132H) vaccine induces a specific antitumour immune response against IDH1(R132H)-mutated tumours in an MHC-humanized animal model. These data underline that mutant IDH may serve as a therapeutic target not only through drug-mediated inhibition of the neomorphic enzymatic function19,20 but also through T-cell-based targeting of the mutant epitope in a disease where the target is an early event in tumorigenesis and hence expressed in all tumour cells, which makes immunological escape unlikely to occur (Supplementary Note 5). Conceptually, patients with low-grade and anaplastic gliomas with a high prevalence of the IDH1(R132H) mutation represent a patient population that may particularly benefit from a tumour vaccine, because these tumours may remain stable or minimally growing for several years but will inevitably recur, often with a more malignant phenotype21, and because there is currently no maintenance therapy available preventing recurrence of this diffusely infiltrating disease in this relatively young and immunologically competent patient population22.

Methods Summary

IDH1 peptide binding to HLA-A*0201 and HLA-DRB1*0101 was analysed by binding prediction algorithms NetMHC and SYFPEITHI, and in vitro T2 and class II REVEAL MHC-peptide binding assays. All animal procedures followed the institutional laboratory animal research guidelines and were approved by the governmental authorities. A2.DR1 and DR4 mice were vaccinated with IDH1(R132H) (p123–142) peptide in complete Freund’s adjuvant (CFA) or Montanide-ISA51 with imiquimod and rmGM-CSF to assess the immunogenicity of IDH1(R132H) in vivo by cytokine analysis of splenocytes. IDH1(R132H)-specific CD4+ T-cell line and clone were generated from splenocytes of vaccinated A2.DR1 mice by re-stimulations with autologous, irradiated and p132–142 (R132H)-loaded splenocytes and limiting dilution for in-depth analysis of IDH1(R132H) T-cell responses. To assess processing and presentation of IDH1(R132H) epitopes, a syngeneic sarcoma cell line was generated by chemical induction, transduced with IDH1(R132H) and wild type, irradiated and used for whole-tumour cell vaccination of A2.DR1 mice. Antitumour activity of induced IDH1(R132H) immune responses was shown by subcutaneous injection of syngeneic IDH1-expressing sarcoma cell line into previously or subsequently vaccinated A2.DR1 mice. Dependence on CD4+ T-cell responses was shown by in vivo CD4 and CD19 depletion. IDH1 wild-type integrity was analysed by enzyme activity measurements in liver and brain. 2-HG production of cells and tumours was analysed enzymatically. Peripheral blood mononuclear cells (PBMCs) from patients with IDH1(R132H) and IDH1 wild-type gliomas were isolated to analyse spontaneous IDH1(R132H)-specific T-cell responses. IFN-γ ELISpot, cytokine ELISA, intracellular flow cytometry, and IFN-γ-secretion assays were used to assess cytokine profile and phenotype of specific T-cell responses after vaccination, in IDH1(R132H) T-cell line and clone, and of spontaneous T-cell responses in glioma patients. IDH1(R132H)-binding IgG was detected with peptide-coated ELISA in serum from vaccinated mice, patients and healthy donors. IDH1(R132H) expression was analysed by immunohistochemistry in human and mouse tumours and by western blot and immunofluorescence in mouse sarcoma cell lines.

Online Methods

Mice

HLA-A*0201 HLA-DRA*0101 HLA-DRB1*0101 transgenic mice devoid of mouse MHC (A2.DR1 mice) C57BL/6-Tg(HLA-DRA*0101,HLA-DRB1*0101)1Dmz-Tg(HLA-A2.1-beta2M)1Bpe-IAbetabtm1Doi-beta2mtm1Doi-H-2Dbtm1Bpe-IAalphatm1Bpe-IEbetatm1Bpe were provided by M. Berard16. HLA-DRB1*0401 transgenic mice (DR4 mice) C57BL/6-Tg(HLA-IEd alpha/HLA-DRB1*0401-I-Ed beta)#39-2 were obtained from Taconic23. NOD/SCID mice were purchased from Charles River. All animal procedures followed the institutional laboratory animal research guidelines and were approved by the governmental authorities. 8–14-week old mice were assigned to age-matched and sex-matched experimental groups.

Patients and patient samples

Patients were diagnosed at the Institute of Neuropathology, University Hospital Heidelberg. Blood samples were taken after patient agreement and approval of the local regulatory authorities. Tissues were obtained from the archives of the Institute of Neuropathology, University Hospital Heidelberg, according to the regulations of the Tissue Bank of the National Center for Tumour Diseases, University Hospital Heidelberg, and used after approval of the local regulatory authorities. IDH1 status was determined by immunohistochemistry at the Institute of Neuropathology, University Hospital Heidelberg. Serum samples of healthy donors were obtained from the Institute for Clinical Transfusion Medicine and Cell Therapy (IKTZ) Heidelberg.

Peptides

Human IDH1 wild type and IDH1(R132H) amino acid sequences IDH1 p118–146 PRLVSGWVKPIIIGRHAYGDQYRATDFVV and PRLVSGWVKPIIIGHHAYGDQYRATDFVV, respectively, cover the amino acid exchange from Arg to His at position 132, including all possible 15-mer IDH1 peptides containing position 132 (Extended Data Fig. 1a), and are identical to mouse sequence except for position 122 (Thr in mouse). Peptide libraries for ex vivo and in vitro stimulation of IDH1 wild type and IDH1(R132H) of 10 and 15-mers (Extended Data Fig. 1b, c) contained the following peptides: IDH1 wild type p123–132: GWVKPIIIGR; IDH1 wild type p124–133: WVKPIIIGRH; IDH1 wild type p125–134: VKPIIIGRHA; IDH1 wild type p126–135: KPIIIGRHAY; IDH1 wild type p127–136: PIIIGRHAYG; IDH1 wild type p128–137: IIIGRHAYGD; IDH1 wild type p129–138: IIGRHAYGDQ; IDH1 wild type p130–139: IGRHAYGDQY; IDH1 wild type p131–140: GRHAYGDQYR; IDH1 wild type p132–141: RHAYGDQYRA; IDH1 wild type p118–132: PRLVSGWVKPIIIGR; IDH1 wild type p120–134: LVSGWVKPIIIGRHA; IDH1 wild type p122–136: SGWVKPIIIGRHAYG; IDH1 wild type p124–138: WVKPIIIGRHAYGDQ; IDH1 wild type p126–140: KPIIIGRHAYGDQYR; IDH1 wild type p128–142: IIIGRHAYGDQYRAT; IDH1 wild type p130–144: IGRHAYGDQYRATDF; IDH1 wild type p132–146: RHAYGDQYRATDFVV; IDH1(R132H) p123–132: GWVKPIIIGH; IDH1(R132H) p124–133: WVKPIIIGHH; IDH1(R132H) p125–134: VKPIIIGHHA; IDH1(R132H) p126–135: KPIIIGHHAY; IDH1(R132H) p127–136: PIIIGHHAYG; IDH1(R132H) p128–137: IIIGHHAYGD; IDH1(R132H) p129–138: IIGHHAYGDQ; IDH1(R132H) p130–139: IGHHAYGDQY; IDH1(R132H) p131–140: GHHAYGDQYR; IDH1(R132H) p132–141: HHAYGDQYRA; IDH1(R132H) p118–132: PRLVSGWVKPIIIGH; IDH1(R132H) p120–134: LVSGWVKPIIIGHHA; IDH1(R132H) p122–136: SGWVKPIIIGHHAYG; IDH1(R132H) p124–138: WVKPIIIGHHAYGDQ; IDH1(R132H) p126–140: KPIIIGHHAYGDQYR; IDH1(R132H) p128–142: IIIGHHAYGDQYRAT; IDH1(R132H) p130–144: IGHHAYGDQYRATDF; IDH1(R132H) p132–146: HHAYGDQYRATDFVV. For mouse vaccination and patient screening, 20-mers IDH1(R132H) p123–142 GWVKPIIIGHHAYGDQYRAT and IDH1 wild type p123–142 GWVKPIIIGRHAYGDQYRAT and 25-mer NY-ESO-1 119–143 PGVLLKEFTVSGNILTIRLTAADHR were used. Negative control peptide for ex vivo and in vitro stimulation was mouse myelin oligodendrocyte glycoprotein (MOG) p35–55 MEVGWYRSPFSRVVHLYRNGK. Negative controls for ELISA establishment were MOG and human immunodeficiency virus (HIV)-1 p17 gag p77–85 SLYNTVATL. In T2 binding assay, human T lymphotropic virus 1 (HTLV1) tax p11–19 LLFGYPVYV was used. MOG peptide was synthesized by Genscript. Tax, HIV, NY-ESO-1 p119–143 and IDH1 10-mer peptides were synthesized in-house; 15- and 20-mers were synthesized by Bachem Distribution Services GmbH and jpt. Peptides were diluted in PBS 10% DMSO at 2.5 mg ml−1 and stored at −80 °C.

MHC binding prediction

IDH1(R132H) p122–142 was subjected to HLA-A*0201 (HLA-A2) peptide binding prediction using NetMHC algorithms24 for 8-mers, 9-mers, 10-mers and 11-mers containing R132H. Peptides with IC50 below 500 nm are defined as weak binders; those with IC50 below 50 nm as strong binders. Prediction for binding of 10-mers to HLA-A2 was also performed using the SYFPEITHI algorithm with positive controls Epstein–Barr virus (EBV) and influenza25.

T2 binding assay

T2 (174 x CEM.T2) cells were obtained from ATCC (CRL-1992) and treated with 5 μg ml−1 β2-microglobulin (BD Pharmingen) and 10 μg ml−1 IDH1(R132H) peptides or HTLV1 tax (p11-19) as positive control peptide for 3 h. Background treatment was with equal volume of 10% DMSO with or without β2-microglobulin. Cells were washed and surface-stained with FITC-conjugated mouse anti-HLA-A*0201 (1 μl per test, BB7.2, BD Pharmingen) and analysed by flow cytometry. Binding values were calculated by background subtraction and relative to positive control.

Class II REVEAL MHC-peptide binding assay

HLA-DRA*0101 HLA-DRB1*0101 (DR1) binding assay was performed by ProImmune (quick check stability assay).

Tumour induction and generation of a syngeneic tumour cell line

500 μg 3-methylcholantrene (MCA) in peanut oil was injected subcutaneously into the flank of A2.DR1 mice. Tumour growth was monitored daily. Tumours reaching 1.5 cm in size were excised, cut into small pieces, embedded in Matrigel (BD Biosciences) and subcutaneously injected into the back of NOD/SCID mice. Tumours were excised, cut into small pieces and cultured in DMEM with 10% FBS (both PAA Laboratories), 50 µg ml−1 gentamycin, 10 mM HEPES, and 50 µM β-mercaptoethanol (all Sigma-Aldrich). When single cells grew out, tissue pieces were removed and adherent cells were cultured in DMEM with 10% FBS, 50 µg ml−1 gentamycin, 10 mM HEPES, and 50 μM β-mercaptoethanol. Cells were embedded in Matrigel, subcutaneously injected into the flank of A2.DR1 mice and tumours were excised for characterization by haematoxylin and staining. Cells were tested for mycoplasma contamination before inoculations.

Stable overexpression of IDH1 and NY-ESO-1

A2.DR1 sarcoma cells were transduced with full-length cDNA of human IDH1(R132H) or IDH1 wild type (NCBI GenBank CR541695.1), or CTAG1B coding for human NY-ESO-1 (NCBI GenBank AJ003149.1), provided by the DKFZ Genomics Facility, in the retroviral vector pMXs-IRES-BsdR (Cell Biolabs, Inc.) using FuGene HD transfection reagent (Promega). Cells were selected with 9 μg ml−1 blasticidin (Sigma-Aldrich) 72 h after transfection. A2.DR1 sarcoma cell clones were generated by limiting dilution and outgrowing clones were screened for IDH1(R132H) expression by immunofluorescence. Clone IVC1 was selected due to high and homogenous IDH1(R132H) expression.

Peptide vaccination and tumour cell inoculation

A2.DR1 or DR4 mice were immunized with 100 μg IDH1(R132H) (p123-142) emulsified in complete Freund’s adjuvant (CFA, Difco) or Montanide-ISA51 (Seppic). CFA or Montanide-ISA51 were emulsified with equal volume of peptide in PBS to 1 mg ml−1 or DMSO in PBS (sham) and mice received 2 subcutaneous injections of 50 μl each into the lateral pectoral regions. For Montanide-vaccination, 300 ng rmGM-CSF (Immunotools) in PBS was subcutaneously injected between injection sites, and Aldara cream containing 5% imiquimod (Meda Pharma) was applied at shaved injection site. For both protocols, mice were boosted after 10 days. No rmGM-CSF was applied at boost. Vaccination only experiments were terminated after 21 days. For tumour cell inoculation experiments, Montanide-ISA51/Aldara was used.

For preventive vaccination, A2.DR1 mice were again boosted on day 21. On day 25, 2 × 106 syngeneic IDH1(R132H)+ or IDH1 wild-type+ sarcoma cells were embedded in Matrigel and subcutaneously injected into the shaved right flank of A2.DR1 mice. For vaccination after inoculation, 5 × 105 syngeneic IDH1(R132H)+ sarcoma cells (clone IVC1) were injected. On day 6, when tumours were detectable in all mice, mice were grouped according to tumour size and vaccinated or sham-treated as above. Mice were boosted on day 13. Vaccination after inoculation against NY-ESO-1 was done by injection of 8 × 105 syngeneic NY-ESO-1+ sarcoma cells and vaccination with NY-ESO-1 (p119-143) or sham-treatment on day 6. Mice were boosted on day 13.

Lymphocyte depletion was done by intraperitoneal injections of anti-CD4 (GK1.5, 1 mg per dose, every 3 days), anti-CD19 (1D3, 500 μg per dose, every 5 days), or corresponding isotype controls (for anti-CD4, LTF-2, 1 mg per dose, every 3 days; for anti-CD19, 2A3, 500 μg per dose, every 5 days; all BioXcell). 8 × 105 syngeneic IDH1(R132H)+ sarcoma cells (clone IVC1) were injected as above. On day 4, when tumours were detectable in all mice, mice were grouped according to tumour size and injected with depleting antibodies (1 group) or isotype controls (2 groups). Mice were vaccinated on day 5 and boosted on day 13 with IDH1(R132H) (p123–142) (depletion groups and one isotype control group) or sham treated (isotype control group). On day 7 and 17, submandibular vein blood was taken from 2 mice per group to verify depletion efficacy by flow cytometry. Tumour growth was measured regularly with a calliper in two dimensions in a blinded fashion and analysed non-blinded.

Tumour cell vaccination

A2.DR1 mice were vaccinated by subcutaneous injection of 107 syngeneic IDH1(R132H)+ (clone IVC1) or IDH1 wild-type+ sarcoma cells irradiated at 100 Gy with 1 μg rmGM-CSF in PBS into the neck. CFA was injected subcutaneously into the lateral pectoral region at both sides (50 μl each). Injection of 107 irradiated sarcoma cells only was repeated 4 times every 10 days and spleens were excised after 47 days.

Isolation of splenocytes and bone marrow cells

Spleens were excised from naive or vaccinated and sham-treated A2.DR1 or vaccinated and sham-treated DR4 mice and bone marrow was isolated from hind limbs of naive A2.DR1 mice. Organs were mashed through a 40 μm cell strainer. Erythrocytes were lysed with ACK buffer, containing 150 mM NH4Cl, 10 mM KHCO3 and 100 μM Na2EDTA, and cells were washed before culturing in RPMI1640 containing 10% FBS and 100 U ml−1 penicillin and 100 μg ml−1 streptomycin (all PAA Laboratories). For splenocytes, medium was supplemented with 1 mM sodium-pyruvate, 2 mM glutamine (both PAA Laboratories), 100 μM non-essential amino acids (Lonza), and 50 μM β-mercaptoethanol (Sigma-Aldrich).

Isolation of tumour-infiltrating lymphocytes

Tumours were excised, washed in HBSS (Sigma-Aldrich) and cut into small pieces before tissue disruption in HBSS supplemented with 0.05% collagenase, 0.1 µg ml−1 TLCK trypsin inhibitor, 10 μg ml−1 DNase I, and 10 mM HEPES, pH 7.4, for 1 h under slow rotation at 37 °C. Dispersed tissue was mashed through a 70 μm cell strainer and lymphocytes were isolated by density-gradient centrifugation using Lympholyte Mouse (Cedarlane).

Generation of antigen-specific T cells

Splenocytes from A2.DR1 mice vaccinated with IDH1(R132H) (p123–142) using Montanide- ISA51/Aldara were stimulated with 10 μg ml−1 IDH1(R132H) (p123–142). After 7 days, medium was exchanged and supplemented with 3% (v/v) ConA supplement (gift from W. Osen) and 15 mM α-methylmonnopyranoside (α-MM, Sigma-Aldrich) weekly. Cells were re-stimulated every 4 weeks with autologous splenocytes from naive A2.DR1 mice irradiated with 30 Gy and loaded with 2 µg ml−1 IDH1(R132H) (p123–142) in ConA supplement and α-MM-containing medium. Cells were used and T-cell clones were generated by limiting dilution (0.5 cells per well) and stimulation with IDH1(R132H) (p123–142)-loaded and irradiated autologous splenocytes after two re-stimulations. Reactive clones were picked according to polarized morphology after 4 weeks and re-stimulated as above.

Isolation of peripheral blood mononuclear cells

Peripheral blood mononuclear cells (PBMC) were isolated from heparin blood of glioma patients by density-gradient centrifugation using lymphocyte separation medium (LSM) 1077 (PAA Laboratories) and cultured in RPMI1640 containing 10% FBS or autologous serum and 100 U ml−1 penicillin and 100 μg ml−1 streptomycin, supplemented with 1 mM sodium-pyruvate, 2 mM glutamine, 100 µM non-essential amino acids, and 50 µM β-mercaptoethanol.

IFN-γ ELISpot

ELISpot white bottom multiwell plates (MAIPSWU10, Millipore) were coated with anti-mouse IFN-γ (AN18, Mabtech) or anti-human IFN-γ (1D1K, Mabtech) and blocked with RPMI1640 containing 10% FBS and 100 U ml−1 penicillin and 100 μg ml−1 streptomycin. Human PBMC or mouse splenocytes were seeded at 5 × 105 per well and stimulated with 20 µg ml−1 or 10 µg ml−1 peptide, respectively. As negative controls, MOG (p35–55) was used at equal concentrations and peptide diluent PBS 10% DMSO (vehicle) was used at equal volume. Positive control was 20 ng ml−1 phorbol myristate acetate (PMA, Sigma-Aldrich) with 1 µg ml−1 ionomycin (Sigma-Aldrich). For T cell line and clone, B cell blasts as antigen-presenting cells (APC) were generated from autologous splenocytes by stimulation with 25 μg ml−1 lipopolysaccharides (LPS, Sigma-Aldrich) and 7 µg ml−1 dextran-sulphate (Amersham Pharmacia Biotech AB) over 3 days, loaded with 0.1 μg ml−1 or 1.0 μg ml−1 IDH1 (p123–142) and control peptides and seeded at 5 × 104 cells per well together with 2.5 × 103 T cells. CD4+ and CD8+ T cells from splenocytes of vaccinated mice were isolated by magnetic cell sorting (MACS) using the pan T cell isolation kit II, mouse, followed by CD4 (L3T4) MicroBeads, mouse (all Miltenyi Biotec). 5 × 103 T cells per well were stimulated with 5 × 104 peptide-loaded (20 μg ml−1) dendritic cells that had been derived from bone marrow cells by treatment with 20 ng ml−1 rmGM-CSF (Preprotech) for 6 days. Peptide presentation was blocked with antibodies (1 μg per well) against HLA-A (W6/32) or HLA-DR (L243, BioLegend). After 36 h, IFN-γ-producing cells were detected with biotinylated anti-mouse IFN-γ (R4-6A2) or anti-human IFN-γ (7-B6-1), streptavidin-ALP (all Mabtech) and ALP colour development buffer (Bio-Rad) and quantified using an ImmunoSpot Analyzer (Cellular Technology Ltd).

Cytokine ELISA

Splenocytes from vaccinated and sham-treated mice were stimulated with 10 µg ml−1 IDH1(R132H) (p123-142), 20 ng ml−1 PMA and 1 µg ml−1 ionomycin as positive control, or peptide diluent PBS 10% DMSO (vehicle) as negative control for 3 days. Supernatants were used on cytokine-specific antibody-coated ELISA plates (Corning) and horseradish-peroxidase (HRP)-based cytokine detection was performed according to manufacturer’s instructions (Ready-SET-Go! ELISA kits, ebioscience) with TMB (ebioscience) and stopped with 1 M H2SO4. OD at 450 nm was measured with an ELISA reader (Thermo Fisher). Cytokine concentrations were calculated using cytokine standard serial dilution. Cytokines detected were IFN-γ, IL-4, IL-6, IL-17 and TNF-α.

IgG ELISA

ELISA plates (Nunc) were coated with IDH1(R132H) and IDH1 wild type (p123–142) for mouse and (p122–136) for patient IgG detection and negative controls MOG (p35–55) and HIV gag (p77–85) (10 µg per well in PBS), washed with PBS 0.05% Tween 20, and blocked with 3% FBS in PBS 0.05% Tween 20. Positive control for patient serum was tetanus toxoid (0.25 μg per well; Merck Chemicals, Ltd). Mouse serum was obtained by puncture of the submandibular vein and centrifugation of whole heparin-mixed blood and diluted 1:50 in blocking solution. Patient and healthy control serum was obtained from serum tubes by centrifugation and used undiluted. Mouse anti-IDH1(R132H) (1:1,000, H09, Dianova) was used as peptide coating control in human and as positive control in mouse serum assays. HRP-conjugated secondary antibodies were sheep anti-mouse IgG (1:5,000, Amersham), goat anti-mouse IgG1, anti-mouse IgG2a, anti-mouse IgG2b, and anti-mouse IgG3 for mouse serum and control and goat anti-human IgG-Fc (all 1:10000, Bethyl Laboratories, Inc.), mouse anti-human IgG1, anti-human IgG2, anti-human IgG3, and anti-human IgG4 (all 1:500, Acris Antibodies, Inc.) for patient serum. Substrate was TMB and reaction was stopped with 1 M H2SO4. OD at 450 nm was measured as above. Data are shown in relation to MOG negative control.

IFN-γ secretion assay

PBMC from glioma patients were stimulated with 40 μg ml−1 IDH1(R132H) (p123–142), MOG (p35–55) as negative, or 1 μg ml−1 staphylococcus-derived enterotoxin B (SEB, Sigma-Aldrich) as positive control and co-stimulated with 2 µg ml−1 anti-human CD28 (CD28.6, ebioscience) in medium containing 10% autologous serum for 16 h. Mouse splenocytes were stimulated with 20 µg ml−1 IDH1(R132H) (p123–142), vehicle as negative control, or SEB as above. Mouse T cells were stimulated with autologous splenocytes loaded with 10 µg ml−1 IDH1(R132H) (p123–142). Positive control was SEB as above. Assay was performed according to manufacturer’s instructions (IFN-γ Secretion Assay Cell Enrichment and Detection Kit (PE), human, 130-054-201; mouse, 130-090-517; Miltenyi Biotec). Briefly, cells were labelled with IFN-γ catch reagent and after secretion period IFN-γ caught on the cell surface was labelled with PE-conjugated antibody. For isolation and enrichment of IFN-γ-secreting cells, cells were incubated with anti-PE microbeads and subjected to magnetic separation. Separated cell populations were stained for flow cytometry. Ex vivo splenocytes were stained and analysed without enrichment.

Flow cytometry

Subsequent to IFN-γ catch assay, selected PBMC populations were blocked with autologous serum and stained with FITC-conjugated mouse anti-human CD3 (SK7), eFluor 450-conjugated mouse anti-human CD4 (RPA-T4), and APC-conjugated mouse anti-human CD8 (SK1, all 1 μl per well, ebioscience). Selected mouse T cells were blocked with rat anti-mouse CD16/32 (93, 0.5 µg per well, ebioscience) and stained with FITC-conjugated rat anti-mouse CD3 (17A2), Pacific Blue-conjugated rat anti-mouse CD4 (RM4-5, both BioLegend), and APC-conjugated rat anti-mouse CD8 (53-6.1, ebioscience; all 0.5 µl/well). Propidium iodide (PI, 1 µg ml−1, Sigma-Aldrich) was used to exclude dead cells. Live single cells were gated on IFN-γ+ CD3+ T cells and analysed for CD4 and CD8 expression.

For IDH1(R132H)-specific T-cell phenotype characterization, splenocytes were ex vivo stimulated with 10 µg ml−1 IDH1(R132H) (p123–142) or vehicle control and T-cell line was stimulated with dendritic cells loaded with 4 μg ml−1 IDH1(R132H) (p123–142) or vehicle control. After 4 days, cells were re-stimulated as before and treated with 5 μg ml−1 brefeldin A (Sigma-Aldrich) for 4.5 h. Cells were blocked as above and stained with Pacific Blue-conjugated rat anti-mouse CD4 and PE-conjugated rat anti-mouse CD25 (PC61, BioLegend; all 0.5 µl per well). For intracellular staining, cells were fixed and permeabilized with cytofix/cytoperm solution and stained in perm/wash buffer (both BD Bioscience) with PE-conjugated rat anti-mouse IFN-γ (XMG1.2), APC-conjugated rat anti-mouse IL-17 (17B7), FITC-conjugated rat anti-mouse IL-4 (BVD6-24G2), and APC-conjugated rat-anti-mouse FoxP3 (FJK-16 s; all 1 μl per well, ebioscience).

For NY-ESO-1-specific T-cell phenotype characterization, splenocytes were ex vivo stimulated with 10 µg ml−1 NY-ESO-1 (p119–143) or MOG as negative control. After 4 days, cells were treated as above, blocked as above and stained with eFluor 450-conjugated rat anti-mouse CD3 (17A2), APC-conjugated rat anti-mouse CD4 (RM4-5), and PE-conjugated rat anti-mouse CD8 (53-6.1, all 0.5 μl per well, ebioscience). Intracellular staining was done as above with FITC-conjugated rat anti-mouse IFN-γ (XMG1.2, 1 µg per well, ebioscience).

Tumour infiltrating lymphocytes were stained for IDH1(R132H)-specific T-helper cells using an HLA-DRB1*01:01 MHC class II tetramer bound to IDH1(R132H) (p123–142) (provided by the National Institute of Health (NIH) tetramer core facility). Cells were blocked as above, stained with APC-conjugated IDH1(R132H)- or CLIP control tetramer (15 μg ml−1, 37 °C, 1 h; NIH) in medium, washed and stained with Pacific Orange-conjugated rat anti-mouse CD45 (1 µl per well, Invitrogen), PE-Cy7-conjugated rat anti-mouse CD3, FITC-conjugated rat anti-mouse CD4, and PE-conjugated rat anti-mouse CD8 (all 1 μl per well, ebiosciences). PI (1 μg ml−1) was used to exclude dead cells. Cells were subsequently gated on CD45+ leukocytes, CD3+ T cells, and CD4+ T cells to analyse tetramer+ T-helper cells.

In all experiments, corresponding isotype controls were used. Cells were measured on a FACS Canto II (BD Biosciences) and data analysis was done using FlowJo software.

Measurement of 2-HG

2-HG production in cells and tumours was analysed as described previously and compared to 2-HG concentrations found in patient tumours, which range from 1 to 30 mM (ref. 26).

Pathological organ analysis

Organs from A2.DR1 mice vaccinated with IDH1(R132H) (p123–142) or sham-treated using Montanide-ISA51/Aldara were excised, fixed with Roti-Histofix 4.5% (Carl Roth GmbH + Co. KG), paraffin-embedded and cut to 3 μm sections which were stained using haematoxylin and eosin and analysed microscopically.

IDH enzymatic activity assay

IDH activity was measured using the isocitrate dehydrogenase activity colorimetric assay kit (BioVision). Liver and brain from A2.DR1 mice therapeutically vaccinated with IDH1(R132H) (p123–142) or sham-treated using Montanide-ISA51/Aldara were cryopreserved, homogenized in IDH assay buffer, and subjected to IDH1/2 activity measurement using NADP+ as cofactor. NADPH production was colorimetrically followed for 30 min after initial 30-min incubation at 37 °C. IDH activity was analysed after subtraction of background without external substrate or NADP+ and calculated according to manufacturer’s instructions using an NADH standard curve.

Western blot

Total protein was isolated by cell or tissue lysis with ice-cold TRIS-HCl, 50 mM, pH 8.0 containing 150 mM NaCl, 1% Nondiet P-40 (Genaxxon Bioscience), 10 mM EDTA, 200 mM dithiothreitol (Carl Roth), 100 μM phenylmethylsulphonyl fluoride (PMSF) and complete EDTA-free (1:50, Roche) for 20 min or in a tissue lyser (Qiagen) using stainless steel beads, respectively, and centrifuged to pellet debris. Protein concentrations were measured via the Bio-Rad protein assay (Bio-Rad) at 595 nm and 30 μg of protein diluted in Laemmli sample buffer were denatured at 95 °C for 5 min and electrophoretically separated on 12% acrylamide-polyacrylamide SDS-containing gels. Proteins were blotted onto nitrocellulose by wet blot at 1.5 mA cm−2 for 1 h. After blocking with 5% milk powder in 0.5 M TBS, pH 7.4, 1.5 M NaCl, 0.05% Tween 20, membranes were incubated consecutively with primary mouse anti-IDH1(R132H) (1:500, H09, Dianova), rat anti-panIDH1 (1:500, W09, Dianova), or mouse anti-NY-ESO-1 (1:500, E978, Sigma-Aldrich) overnight at 4 °C, and mouse anti-α-tubulin (1:5,000, Sigma-Aldrich) as loading control for 1 h at room temperature. Staining with secondary HRP-conjugated anti-rat (1:(1000xF), Dako) or anti-mouse (1:5000, GE Healthcare) was performed at room temperature for 1 h, followed by chemiluminescent development using ECL or ECL prime (both Amersham).

Immunofluorescence

Cells were seeded on glass coverslips, grown until 70–90% confluent and fixed and permeabilized with Cytofixx Pump Spray (Cell Path) at −20 °C and subsequent 4% PFA. For blocking and staining, 5% FBS in PBS was used. Primary antibodies were mouse anti-IDH1(R132H) (1:50, H09, Dianova) and mouse anti-NY-ESO-1 (1:100, E978, Sigma-Aldrich) and secondary antibody was donkey anti-mouse AlexaFluor 488 (1:300, Molecular Probes, Invitrogen). Vectashield HardSet Mounting Medium with DAPI (Vector Laboratories) was used for mounting and nuclear staining. Images were taken on DM IRB microscope (Leica).

Immunohistochemistry

A2.DR1 mouse tumour tissue was fixed with Roti-Histofix 4.5%, paraffin-embedded and cut to 3 μm sections and processed using a Ventana BenchMark XT immunostainer. Tissue was deparaffinized with HistoClearTMII and rehydrated. Antigen retrieval was performed using Cell Conditioning Solution CC1 (Ventana Medical Systems, Inc.) for 30 min. Endogenous peroxidase activity was blocked by 3% hydrogen peroxide in PBS for 10 min. Blocking was performed using 5% FBS in PBS for 30 min. N-Histofine mouse stain kit (Nichirei Biosciences, Inc.) was used to reduce unspecific binding of monoclonal mouse anti-IDH1(R132H) (1:50, H09, Dianova) and rat anti-CD3 (1:300, Dako) in mouse tissue according to manufacturer’s instructions. Briefly, before adding primary antibody, endogenous mouse tissue IgG was blocked using reagent A and was blocked with reagent B before incubation with Simple Stain Mouse MAX PO. Counterstaining was performed with hemalum (Carl Roth GmbH + Co. KG) for 2 min. Stained tissue was dehydrated and mounted with Histomount (National Diagnostics). Haematoxylin and eosin staining was performed using haematoxylin and bluing reagent for 4 min.

Image analysis

20 images per tumour were randomly taken of the tumour bulk with 20× objective using a LEICA DM LB2 microscope. IDH1(R132H)+ A2.DR1 sarcoma cells and CD3+ lymphocytes were counted by ImageJ with an algorithm using background subtraction and colour deconvolution plugin. IDH1(R132H)+ and CD3+ cell count of 20 images was set in relation to calculated tumour volume in mm3.

HLA typing

Genomic DNA was isolated from patient heparin blood using the QIAamp DNA Blood Mini Kit (Qiagen). PCR-based typing was performed using HLA-A and HLA-DR type-specific primer pairs lyophilized in a 96-well plate (HLA-A* CTS-PCR-SSP Minitray Kit and HLA-DRB1* CTS-PCR-SSP Minitray Kit) and Mastermix 5.0% for HLA-DRB1* and 7.5% for HLA-A* (all from Department of Transplantation Immunology, University Clinic Heidelberg) with Taq-polymerase (Fermentas). PCR was performed according to manufacturer’s instructions, products were separated on a 1.5% agarose gel containing GelRed (1:10,000; Genaxxon bioscience) and analysis was done according to manufacturer’s instructions.

Statistics

The mean of variables was compared with Student‘s t-test, assuming unequal variances, or Welch‘s ANOVA, when more than two groups were compared. In that case, a pairwise t-test with Bonferroni correction was performed afterwards. The Wilcoxon rank-sum test was used to compare the median of variables. Data shown in Fig. 3d were log2 transformed before analysis. The area under the curve (AUC) of tumour area over time was calculated through the MIfuns package, and median AUC was compared with the Wilcoxon rank-sum test. Odds were compared with Fisher‘s exact test. The Bonferroni method was applied to adjust for type I error inflation due to multiple testing and P < 0.05 was considered statistically significant (http://www.R-project.org/). All tests were two-sided. Analyses were carried out using R version 2.15.2. Sample sizes for animal experiments were empirically chosen. All in vitro assays were performed three times; patients were tested once except ELISpot responders who were tested at least twice; mouse experiments were repeated five times for peptide vaccination, three times for peptide vaccination with tumour challenge, and twice for whole tumour cell vaccination. Peptide vaccination with lymphocyte depletion and tumour challenge was performed once.