D-2-Hydroxyglutarate Is an Intercellular Mediator in IDH-Mutant Gliomas Inhibiting Complement and T Cells

Site-specific mutations of isocitrate dehydrogenase (IDH)-1 or -2 are present in 80% to 90% of patients with diffuse WHO grade II–III gliomas and a small subset of patients with WHO grade IV glioblastomas (1–4). IDH mutations are also present but less penetrant in acute myeloid leukemia (5), angioimmunoblastic T-cell lymphoma (6), and chondrosarcomas (7). However, precisely how IDH mutations might confer an advantage to tumorigenesis is not well understood.

Missense mutations of R132 in IDH-1 or R172 in IDH-2 that change this arginyl residue to any of several other residues confer a remarkable gain of function to IDH catalytic activity enabling mutant enzyme to stereospecifically reduce isocitrate to D-2-hydroxyglutarate (D 2-HG; refs. 2, 8, 9) rather than its normal product α-ketoglutarate. D 2-HG accumulates to 30 mmol/L within (10) and 3 mmol/L surrounding (11) gliomas carrying a mutant IDH-1 or IDH-2 gene. D 2-HG alters tumor cell metabolism and epigenetic regulation (12–14), but the full significance of IDH mutations or more precisely the unique nature of excessive D 2-HG accumulation is undefined. For instance, we now know that tumor IDH mutation tightly correlates to the absence of microthrombi within the tumor vasculature of diffuse gliomas, and that D 2-HG directly suppresses ex vivo activation and thrombosis of purified platelets (15). Potentially, then, tumor-derived D 2-HG functions as an intercellular mediator that affects nonneoplastic cells of the tumor microenvironment. Tumor-infiltrating CD4⁺ helper and CD8⁺ cytotoxic T cells are present in the glioma microenvironment (16), and mutant IDH associates with fewer infiltrating immune cells, including macrophages, T cells, and B cells, in tumors (17–19), and IDH-mutant gliomas may escape from natural killer (NK) cell immune surveillance by downregulation of their NK group 2, member D (NKG2D) ligand expression (20).

Complement is a key component of the innate immune system that defends against pathogen invasion and clears apoptotic cells and immune complexes. When activated by either classical, alternative, or lectin pathways, activated complement forms membrane attack complex (MAC) pores that lyse targeted cells (21). Complement activation also leads the deposition of C3b(iC3b) fragments on target cells for “opsonization” that facilitates phagocytosis through interactions with C3b(iC3b) receptors (C3aR) expressed on phagocytes. Recent studies (22–24) also found that complement directly regulates T-cell function, in part through signaling of G-protein coupled C3aR and C5aR receptors on antigen-presenting cells and T cells.

Here, we determined whether the immunologic microenvironment of adult diffuse gliomas is affected by IDH mutational status. We find that IDH mutation associates with reduced complement activation, decreased CD4⁺, FOXP3⁺, and CD8⁺ T-cell infiltration in gliomas in situ, and that D 2-HG directly suppresses these essential elements of both innate and adaptive immunity.

Expanded Materials and Methods are presented in a supplement to this article.

Tissues were obtained from patients diagnosed with primary high-grade astrocytoma between 1997 and 2017. All tumor samples were classified or reclassified according to the WHO Classification 2016 (25). Patients underwent initial surgery at the Department of Neurosurgery, Odense University Hospital, Denmark, or at the Department of Neurosurgery, Heinrich Heine University, Düsseldorf, Germany. None of the patients had received treatment prior to surgery. Of the 72 patients included in the current study, 23 were WHO grade III anaplastic astrocytomas and IDH-mutant (mIDH), 16 were WHO grade III anaplastic astrocytomas and IDH-wild-type (wtIDH), 14 were WHO grade IV glioblastomas with mIDH, and 19 were WHO grade IV glioblastomas with wtIDH. IDH status was determined by immunohistochemistry using an antibody against the most common IDH-1-R132H mutation (clone H14, Dianova) using the BenchMark Ultra IHC/ISH staining system (Ventana Medical Systems, Inc.; ref. 26), and/or by next-generation sequencing as previously described (27). Of the 37 detected IDH mutations, 31 were IDH-1-R132H, three were IDH-1-R132C, and one each corresponded to IDH-1-R132S, IDH-1-R132G, or IDH-2 R140W.

Additionally, double immunohistochemistry with antibodies against C3/C3b and the tumor marker oligodendrocyte transcription factor (OLIG2) was performed on six of the 72 astrocytomas included in the patient cohort (one mIDH and one wtIDH anaplastic astrocytoma, two mIDH and two wtDH glioblastomas) to verify and localize deposition of C3 on tumor cells.

The potential effects of D 2-HG in inhibiting the classical and alternative pathways of complement activation were analyzed using antibody-sensitized sheep erythocytes (E^shA) or rabbit erythrocytes (E^rabb) following well-established protocols (28).

Complement convertases of the classical and alternative pathways were analyzed following a published protocol using E^shA or E^rabb (29, 30).

Complement-mediated brain tumor cell damage assay was done based on the measurement of lactate dehydrogenase (LDH) leakage using a commercial kit (Sigma-Aldrich).

E^shA were incubated with 2% C5-depleted serum in gelatin veronal buffer with calcium and magnesium (GVB⁺⁺) containing defined concentrations of D 2-HG. For negative controls, 5 mmol/L EDTA was added to the buffer. After 10 minutes at 37°C, E^shA were washed and stained with an Alexa Fluor 488-conjugated anti-human C3 antibody (MP Biomedicals) for additional 30 minutes on ice, followed by flow cytometry analysis.

The myeloid cell line U937 was differentiated into macrophages for the complement opsonization-mediated phagocytosis assay based on a published protocol (31, 32).

Nylon wool-enriched T cells, or negative selection-purified CD4⁺ and CD8⁺ T cells from WT mice were activated by monoclonal antibodies against CD3 and CD28, then cultured in different polarization conditions in the presence of different concentrations of D 2-HG. The inhibitory effect of D 2-HG was assessed by measuring the proliferation of the activated T cells using carboxyfluorescein succinimidyl ester (CFSE) dilution and bromodeoxyuridine (BrdU) incorporation. In addition, cytokines produced by the activated T cells were quantitatively assessed in the culture supernatants by ELISA, and the generation of Tregs were assessed by analyzing CD4⁺ CD25⁺ FOXP3⁺ cells using flow cytometry.

Impact of D 2-HG on T-cell migration was assessed in a conventional transwell migration assay.

Dendritic cells (DC) were generated from bone marrow using a published protocol (33), and their function was assessed using antigen-specific T cells from ovalbumin peptide 323–339 (OVA_323-339)-specific TCR transgenic mice (OT II mice) and ovalbumin peptide 257-264 (OVA_257-264)-specific TCR transgenic mice (OT I mice).

Statistical analyses were performed in GraphPad Prism (Version 5). Mann–Whitney U test or Student unpaired t test, as appropriate, were used to investigate the difference in protein expression between mIDH and wtIDH tumors. One-way ANOVA with Bonferroni correction was used to analyze data of more than two groups and Student t test was used to analyze data of two sets. P < 0.05 was considered significant.

To examine whether IDH mutational status, and thus the presence of excessive D 2-HG, associates with complement activation in the tumor microenvironment, we performed immunohistochemistry on tissue samples from 72 patients using an antibody against C3(C3b) fragments deposited on cell surfaces after activation of the complement cascade. We found in representative sections (Fig. 1A and B) that deposition of these complement fragments, assessed by their staining density, was less in tumors from the 37 patients with mIDH than in samples from 35 patients with wtIDH. Accordingly, the overall intensity (Fig. 1C; P < 0.001) and fraction score (Fig. 1D; P < 0.001) of C3(C3b) were lower in WHO grade III and IV astrocytomas with mIDH as compared with WHO grade III and IV astrocytomas with wtIDH (Supplementary Table S1). Similar results were found when separately analyzing C3(C3b) immunopositivity in the groups of anaplastic astrocytomas and glioblastomas (Supplementary Table S1). Further, we found that mIDH astrocytomas tended to have less complement deposition on the luminal surfaces of small blood vessel and capillaries compared with wtIDH tumors (P = 0.085), and this was especially the case when looking separately at glioblastomas (P < 0.01; Fig. 1E–G; Supplementary Table S1). Additionally, the intensity of deposited C3(C3b) in necrotic zones was lower in mIDH glioblastomas than in wtIDH glioblastomas (Fig. 1E, F, H, and Supplementary Table S1). Double labeling with C3(C3b) and the tumor marker OLIG2 showed deposition of complement fragments in close proximity to OLIG2⁺ nuclei suggesting that C3(C3b) is also deposited on tumor cell surfaces. This was seen both in astrocytomas with wtIDH and mIDH (Fig. 1I and J).

To explore the mechanism underlying the reduced complement activation/deposition in gliomas with IDH mutation, we used a conventional complement-mediated hemolytic assay to test whether D 2-HG inhibits the classical pathway of complement activation and the cellular lysis it generates through MAC complex formation (28). This experiment showed D 2-HG suppressed MAC-induced hemolysis, that suppression was a function of the concentration of normal human serum (NHS) and hence MAC complex abundance, and that this inhibition became statistically significant by 5% NHS (Fig. 2A). We also tested the effects of different concentrations of D 2-HG in this assay, using a fixed amount of NHS, to find that D 2-HG inhibited the classical pathway of complement activation in a dose-dependent manner with a minimally effective concentration of 2 mmol/L (Fig. 2B).

D 2-HG could inhibit complement-mediated cell damage at multiple steps of the complement activation cascade. To explore underlying mechanisms, we determined whether D 2-HG affected assembly of C3 and C5 convertases leading to MAC complex-induced hemolysis (29). These assays showed that D 2-HG had no effect on C3 convertase assembly in the classical pathway of complement activation (Fig. 2C) but significantly inhibited the assembly of C5 convertases of the classical pathway at concentrations of 20 mmol/L and 30 mmol/L (Fig. 2D).

To determine whether D 2-HG inhibits the activities of preassembled C3 and/or C5 convertases in the classical pathway of complement activation, we incubated E^shA with C3- or C5-depleted serum alone to allow the assembly of C3 or C5 convertases and, after washing, we incubated these cells with guinea pig serum in the presence of EDTA (to prevent new convertase assembly) and different concentrations of D 2-HG to measure complement-mediated hemolysis (29). These experiments showed that D 2-HG modestly decreased complement-mediated hemolysis in a dose-dependent manner (Fig. 2E and F), indicating that D 2-HG can inhibit the activity of both assembled C3 and C5 convertases in the classical pathway of complement activation, and that D 2-HG is not inhibiting complement activity through simple Ca⁺⁺ ligation (15).

The alternative complement pathway is a distinct, major route for complement activation that additionally amplifies complement activation initiated through other pathways. To evaluate the potential effects of D 2-HG on this route to complement activation, we used an E^rabb-based complement-mediated hemolytic assay (28). The results of this assay showed that D 2-HG additionally inhibited the alternative pathway of complement activation (Fig. 2G). We then incubated E^rabb with 20% NHS in the absence or presence of different concentrations of D 2-HG over the range from 0.1 to 30 mmol/L, and found that D 2-HG also significantly reduced the complement-mediated hemolysis in a dose-dependent manner with a minimally required concentration of 5 mmol/L (Fig. 2H).

To elucidate the mechanisms by which D 2-HG inhibits the alternative pathway of complement activation, we used convertase assays similar to the above-described protocol for the classical pathway, but with the E^rabb as the complement activator (28, 29). Because both C3 and C5 convertases in the alternative pathway require C3b, these two enzymes cannot be distinguished by using C3- or C5-depleted sera. These experiments showed that at concentrations of 20 mmol/L and 30 mmol/L, D 2-HG significantly inhibited the assembly of the C3/C5 convertases of the alternative pathway (Fig. 2I).

We next tested the effects of D 2-HG on preassembled C3/C5 convertases in the alternative pathway of complement activation by incubating E^rabb with C5-depleted serum, washing the cells, then incubating them again with guinea pig serum in the presence of EDTA and varied amounts of D 2-HG. These experiments showed that D 2-HG did not significantly inhibit the activity of preassembled C3/C5 convertases in the alternative pathway of complement activation (Fig. 2J), so there are enzymatic and functional differences between preassembled and assembled convertases.

The above studies based on complement-mediated hemolysis assays suggest that D 2-HG could inhibit complement activation and thereby MAC-mediated brain tumor cell damage. To test this, we incubated antibody-sensitized T98 glioblastoma cells with different concentrations of complement in the presence of varied D 2-HG concentrations, then evaluated the complement-mediated cell injury by measuring levels of LDH that leaked from the cells. These experiments showed that D 2-HG significantly inhibited cell injury from complement-mediated cellular damage in a dose-dependent manner (Fig. 2K).

In addition to the MAC formation, complement activation deposits C3b(iC3b) on target cells for opsonization that facilitates phagocytosis (34). To determine whether this complement function also was compromised by D 2-HG, we examined the effects of D 2-HG on both C3b(iC3b) opsonization and complement-mediated phagocytosis. We incubated E^shA with C5-depleted serum (to avoid MAC formation and cellular lysis) in the presence of different concentrations of D 2-HG, then quantitated the levels of C3b(iC3b) deposited on the cell surface by flow cytometry. We found that C3 deposition on the cells was significantly decreased by D 2-HG in a dose-dependent manner (Fig. 3A and B). In parallel experiments to assess phagocytosis, we first incubated E^shA with C5-depleted serum in the absence or presence of 30 mmol/L D 2-HG, then fluorescently labeled them before mixing these cells with macrophages labeled with a different fluorophore. After incubation for either 30 or 120 minutes, phagocytosis was quantitated by analyzing the double-positive cells by flow cytometry (Fig. 3C and D). This showed that D 2-HG markedly reduced the efficiency of complement-mediated phagocytosis.

To elucidate a potential association between IDH mutations and the adaptive immune system in patients with WHO grade III and IV astrocytomas, we quantitatively evaluated tumor-infiltrating lymphocytes by immunohistochemistry in the same 72 primary tumors we used to assess complement deposition (Fig. 1). We found fewer tumor-infiltrating CD4⁺ T cells in the group of mIDH gliomas than in the group of wtIDH gliomas of WHO grade III and IV (P < 0.01; Fig. 4A–C). Similar outcomes were obtained when analyzing the subgroup of WHO grade III anaplastic astrocytomas (P < 0.05) stratified according to IDH mutational status, with a similar tendency in subgroup of WHO grade IV glioblastomas (P = 0.14; Supplementary Table S1). We then investigated the infiltration of glioma tissues by CD8⁺ cytotoxic T cells and observed decreased numbers of these cells in mIDH compared with wtIDH WHO grade III or IV gliomas (P < 0.001; Fig. 4D–F). Again, this difference was also observed for the subgroup of anaplastic astrocytomas (P < 0.01), with a similar tendency in the subgroup of glioblastomas that was not, however, significant (P = 0.33). Interestingly, staining for the Treg cell marker FOXP3 showed that numbers of tumor-infiltrating FOXP3⁺ T cells were lower in mIDH grade III and IV gliomas compared with their wtIDH counterparts (Fig. 4G–I; P < 0.01). Although both FOXP3 positive and negative cells were decreased, the difference in the tumor-infiltrating FOXP3⁺/CD4⁺ ratio between the two groups was not statistically significant (P = 0.13; Fig. 4J).

Although D 2-HG inhibits CD8⁺ T-cell accumulation in tumors (17), whether it directly inhibits proliferation of activated T cells and/or stimulated cytokine production is unknown. We labeled purified T cells with fluorescent CFSE and activated them with anti-CD3 and anti-CD28 monoclonal antibodies in the absence or presence of D 2-HG at concentrations ranging from 5 to 30 mmol/L. We then assessed proliferation of these activated T cells by CFSE dilution (Fig. 5A) to find D 2-HG significantly inhibited the proliferation of activated T cells. We confirmed this by assessing BrdUrd incorporation into the total DNA content of dividing cells (Fig. 5B). We also found that D 2-HG suppressed IFNγ production from the activated T cells in a dose-dependent manner (Fig. 5C).

To test whether D 2-HG inhibits different subsets of effector CD4⁺ T cells, we isolated CD4⁺ T cells from naïve WT mice, activated them by monoclonal antibodies against CD3 and CD28, and cultured these cells in Th1, Th17, and Treg polarization conditions in the presence or absence of D 2-HG. We then assessed the proliferation of these various cells to find D 2-HG significantly inhibited proliferation of activated Th1, Th17, and Tregs (Fig. 5D). Surprisingly, when we analyzed the differentiation of CD4⁺CD25⁺FOX3⁺ Tregs in the presence of D 2-HG, we found that 30 mmol/L D 2-HG significantly augmented differentiation of these cells (Fig. 5E and F). However, levels of IL10 in the Treg culture supernatants were reduced in the presence of 30 mmol/L D 2-HG. Thus, while D 2-HG augments Treg differentiation, it concurrently inhibits proliferation of the differentiated Tregs (Fig. 5D). Accordingly, D 2-HG led to reduced numbers of tumor-infiltrating Tregs (Fig. 4G–I) in vivo and decreased production of IL10 in vitro (Fig. 5G).

Inhibition of T-cell proliferation by D 2-HG would reduce the presence of both CD4⁺ and CD8⁺ T cells in tumors from patients with mutant IDH, as would suppressed T-cell migration. To address this question, we set up a transwell T-cell migration assay following an established protocol using Chemokine (C-C motif) ligand 19 (CCL19) as a chemoattractant to evaluate the effect of D 2-HG on T-cell migration. These studies showed that 30 mmol/L D 2-HG significantly inhibited the migration of both CD4⁺ and CD8⁺ T cells (Fig. 5H).

DCs are pivotal for T-cell activation, so DCs were differentiated from bone marrow cells in the presence of defined concentrations of D 2-HG and then compared 6 days later with the DCs differentiated in the absence of D 2-HG using cell surface markers CD11c, MHCII, CD11b, CD80, and CD86. We found that D 2-HG, even at 30 mmol/L, did not significantly affect the ratio of differentiated DCs (Fig. 6A and B), suggesting that D 2-HG does not indirectly suppress T-cell function through DC differentiation.

Even though D 2-HG did not have an appreciable effect on DC differentiation, it might still affect their function. We assessed this in the presence or absence of varied concentrations of D 2-HG by mixing the same numbers of these DCs with purified T cells from OT II transgenic mice that specifically recognize the ovalbumin 323-339 (OVA_323-339) peptide together with this peptide antigen. We then compared proliferation of the OT II T cells activated by the DCs presenting the OVA peptide, again by BrdUrd incorporation, and quantified their functional response by IFNγ ELISA. These assays showed that there was no appreciable difference between DCs differentiated in the absence or presence of D 2-HG in stimulating antigen-specific T-cell proliferation (Fig. 6C) or IFNγ production (Fig. 6D).

To determine whether D 2-HG affects DC antigen processing or presentation and to confirm that D 2-HG inhibits antigen-specific CD4⁺ and CD8⁺ T cells, we first incubated DCs with OVA protein in the presence or absence of 30 mmol/L D 2-HG for 4 hours, then washed the cells and incubated them with CD4⁺ T cells from OT II mice or CD8⁺ T cells from OT I mice for another 3 days. We then analyzed proliferation of the OVA-specific CD4⁺ T cells and CD8⁺ T cells by flow cytometry, and measured levels of IFNγ produced from the activated CD4⁺ T cells, and levels of granzyme B produced from the activated CD8⁺ T cells in the culture supernatants by ELISA. These assays showed that both OVA-specific CD4⁺ and CD8⁺ T cells had comparable proliferation (Fig. 6E) and produced similar levels of IFNγ (CD4⁺; Fig. 6F) or granzyme B (CD8⁺; Fig. 6G) after coculturing with DCs incubated with OVA protein in the presence or absence of D 2-HG. This indicates that D 2-HG even at 30 mmol/L has no effect on DC antigen processing or presentation. In contrast, when the same DCs with processed/presented OVA (in the absence of D 2-HG) were incubated with CD4⁺ T cells from OT II mice or CD8⁺ T cells from OT I mice with or without 30 mmol/L D 2-HG, proliferation of the OVA-specific CD4⁺ and CD8⁺ were significantly inhibited by D 2-HG (Fig. 6E), as was their production of IFNγ (CD4⁺; Fig. 6F) or granzyme B (CD8⁺) (Fig. 6G). This shows that D 2-HG directly inhibits antigen-specific CD4⁺ and CD8⁺ T cells without interfering with DC antigen processing or presentation.

We report that IDH mutation associates with significantly decreased levels of complement deposition within human glioma tissues. From a molecular perspective, we found that D 2-HG inhibited both the classical and the alternative pathways of complement activation, reduced MAC-mediated cellular injury, and decreased complement-mediated opsonization and phagocytosis. We found that IDH mutation was also significantly associated with reduced numbers of tumor-infiltrating CD4⁺, CD8⁺, and FOXP3⁺ T cells in tumor tissue samples from patients with either WHO grade III anaplastic astrocytomas or WHO grade IV glioblastomas. In mechanistic studies, we found that although D 2-HG did not inhibit the differentiation of DCs or their function after differentiation, D 2-HG directly suppressed proliferation of activated T cells and their production of key cytokines. These results elucidate a novel transcellular effect of tumor-derived D 2-HG on select cells and effector pathways of the immune system in a tumor microenvironment.

Complement forms a central part of the host immune surveillance mechanism against tumor cells, yet may have opposing roles in tumorigenesis because activated complement also promotes inflammation that favors tumor growth. Complement can be activated on tumor cells, either directly by the tumor cells themselves (35, 36) or by tumor-reactive antibodies that bind to neoantigens on the tumor cell surface, enabling MAC-mediated lysis and facilitated phagocytosis to dissolve the tumor (37). Conversely, tumors significantly upregulate expression of complement inhibitors including CD55 and CD59 on their surface that shield them from complement-driven attacks (38, 39). Our finding that D 2-HG significantly inhibited the activation of complement from both the classical and the alternative pathways suggest a new mechanism that would facilitate mIDH glioma cell survival. The classical pathway of complement activation is primarily initiated by antibody-antigen complexes. During tumor development, the proteome of the transformed cell includes neoantigens that can provoke B-cell response to generate tumor-reactive antibodies against these neoepitopes. These antitumor antibodies, once bound to their antigens on tumor cell surface, initiate selective complement activation by binding circulating C1 molecules. This leads to MACs formation of transmembrane pores that permeabilize tumor cells, and additionally deposits C3b(iC3b) on the tumor cell surface to facilitate subsequent clearance by phagocytosis. The effect of the inhibition of the classical pathway activation of complement by D 2-HG would thus act to reduce the efficiency of antitumor antibodies in at least two ways.

In addition to direct attack by antitumor antibodies and their activation of complement through the classical pathway, altered expression patterns of surface molecules on tumor cells can trigger complement activation through the alternative pathway. This mechanism also forms MAC and deposits C3b(iC3b) on the target cells. Furthermore, the alternative pathway is a component of the amplification loop for complement activation initiated by other pathways, including the classical pathway (40). Complement is highly conserved among species, and we observed D 2-HG inhibited complement from normal mice, rats, and guinea pigs (not shown). Thus, a propensity of D 2-HG to enhance tumorigenesis, in part, proceeds through effects on different routes to complement activation.

Our studies suggest that D 2-HG does not have a significant effect on at least the C3 convertases from the classical pathway of complement activation (Fig. 2). However, detection of deposited C3 antigens (C3b/iC3b) is robust. This suggests that alternative pathway might play a major role in activating complement in these tumors. In addition, our C3b uptake assays (Fig. 3) also suggest that even though D 2-HG might not have a drastic effect on the C3 convertases, it significantly inhibits the binding (deposition) of activated C3 onto the cell surface.

Inhibition of complement convertases by D 2-HG will be independent of Ca⁺⁺ sequestration (15), because extracellular Ca⁺⁺ significantly exceeds D 2-HG, although suppression of T-cell function might reflect effective Ca⁺⁺/D 2-HG ligation at the far lower intracellular Ca⁺⁺ levels (15). These new data also suggest that D 2-HG from the altered catalytic activity of mutant IDH in tumor cells has a profound role in suppressing both the innate and the adaptive immune systems that may underlie reduction of tumor-infiltrating T cells in human astrocytoma.

The adaptive immune system plays a vital role in tumor immune surveillance (41, 42) through tumor-reactive T cells, activated by tumor antigen-presenting cells such as DCs, to proliferate, release cytotoxins such as granzymes, and produce inflammatory cytokines including IFNγ. T cells also facilitate the humoral response to produce tumor-directed antibodies that activate complement on tumor cells leading to the assembly of MAC pores, lysis, and recognition and engulfment by macrocytic cells. Contravening this, tumor cells stimulate induction of myeloid-derived suppressor cells, upregulate programmed death-ligand 1 (PD-L1; ref. 43) on their cell surface (44), and as we show here interfere with complement activation, and directly suppress T-cell function.

We found that astrocytic gliomas expressing either mutant IDH-1 or IDH-2 contained significantly fewer tumor-infiltrating T cells relative to histologically similar tumors with WT IDH. Importantly, we show that the abundance of CD4⁺ helper cells, cytotoxic CD8⁺ T cells, and total FOXP3⁺ Tregs was lower in gliomas with mutant as compared with tumors with WT IDH. Previously, the abundance of CD8⁺ cells in a smaller cohort glioblastoma patients was found to be reduced in tumors with mutant enzyme relative to tumors with WT enzyme (17), and we confirmed a reduction of these cells, but we also found fewer CD4⁺ and FOXP3⁺ T cells in a larger number of both WHO grade III and grade IV tumors. These data indicate that tumors expressing mutant IDH and synthesizing D 2-HG would be subject to lower levels of immune surveillance and immune-mediated elimination that includes reduced NK cell-mediated immunosurveillance (20).

The fact that any of the several mutations of IDH that induce the gain-of-function production of D 2-HG were associated with fewer tumor-infiltrating T cells suggests that the D 2-HG product itself is the likely functional effector limiting immunosurveillance. In a recent study (17), D 2-HG was found to suppress STAT1 activation and CD8⁺ T-cell trafficking into gliomas correlating to loss of NK cell ligand expression (20). Additionally, we found that D 2-HG inhibited the proliferation of activated T cells and their cytokine production, which are central components of acquired immunity. However, in contrast to suppression of the proliferation of activated T cells and their production of cytokines, D 2-HG did not have an appreciable effect on DC differentiation or function, while it actually stimulated FOXP3⁺ CD4 T-cell proliferation, although this occurred with sharp reduction of their stimulated function. This finding indicates that D 2-HG is selective in the cells and processes it inhibits and is not a general cytotoxin or cell-cycle inhibitor.

The limitations of this study include that we postulate, but do not test, whether glioblastomas expressing IDH mutations abundantly release D 2-HG to their environment, similar to their extensive release of glutamate. The immune-inhibition we explore here occurs at the high D 2-HG levels found within or within centimeters of gliomas, but because the extracellular D 2-HG concentration in these locations is only modeled, we do not know the actual concentrations of D 2-HG experienced by tumor-infiltrating lymphocytes. Additionally, this study correlates IDH mutational status in human glial tumors with reduced immune cell infiltration but did not directly test the role of the IDH mutation in isolation using murine xenograft models.

Overall, our studies found that the overproduction of D 2-HG in tumors expressing mutant IDH-1 and IDH-2 influences the tumor microenvironment by intervention in immunosurveillance at two key points, extracellular suppression of both classical and alternative complement deposition, as well as direct suppression of the T-cell response. These results provide new insights into the mechanism by which the oncometabolite D 2-HG facilitates tumorigenesis of glioma cells carrying the IDH mutations.

No potential conflicts of interest were disclosed.

Conception and design: B.W. Kristensen, T.M. McIntyre, F. Lin

Development of methodology: L. Zhang, M. Sorensen, B.W. Kristensen, T.M. McIntyre

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Zhang, M. Sorensen, B.W. Kristensen, G. Reifenberger

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Zhang, M. Sorensen, B.W. Kristensen, F. Lin

Writing, review, and/or revision of the manuscript: L. Zhang, M. Sorensen, B.W. Kristensen, G. Reifenberger, T.M. McIntyre, F. Lin

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Zhang, B.W. Kristensen

Study supervision: B.W. Kristensen, F. Lin

This work is supported in part by grants NIH R01 DK 10358 (F. Lin), Cleveland Clinic Center of Excellence in Cancer-Associated Thrombosis Award (F. Lin and T.M. McIntyre), and VeloSano Pilot Project Award (T.M. McIntyre and F. Lin). We thank Justin D. Lathia for his helpful insights and discussions in creating and formulating this article. We also thank technician Helle Wohlleben and senior histotechnician, project coordinator Ole Nielsen for assistance with immunohistochemical staining.

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.