Reader TTS

ABSTRACT The dynamics and phenotypes of intratumoral myeloid cells during tumor progression are poorly understood. Here we define myeloid cellular states in gliomas by longitudinal

single-cell profiling and demonstrate their strict control by the tumor genotype: in isocitrate dehydrogenase (IDH)-mutant tumors, differentiation of infiltrating myeloid cells is blocked,

resulting in an immature phenotype. In late-stage gliomas, monocyte-derived macrophages drive tolerogenic alignment of the microenvironment, thus preventing T cell response. We define the

IDH-dependent tumor education of infiltrating macrophages to be causally related to a complex re-orchestration of tryptophan metabolism, resulting in activation of the aryl hydrocarbon

receptor. We further show that the altered metabolism of IDH-mutant gliomas maintains this axis in bystander cells and that pharmacological inhibition of tryptophan metabolism can reverse

immunosuppression. In conclusion, we provide evidence of a glioma genotype-dependent intratumoral network of resident and recruited myeloid cells and identify tryptophan metabolism as a

target for immunotherapy of IDH-mutant tumors. SIMILAR CONTENT BEING VIEWED BY OTHERS IMMUNE LANDSCAPE OF ONCOHISTONE-MUTANT GLIOMAS REVEALS DIVERSE MYELOID POPULATIONS AND TUMOR-PROMOTING

FUNCTION Article Open access 05 September 2024 MTAP LOSS CORRELATES WITH AN IMMUNOSUPPRESSIVE PROFILE IN GBM AND ITS SUBSTRATE MTA STIMULATES ALTERNATIVE MACROPHAGE POLARIZATION Article Open

access 09 March 2022 QUINOLINATE PROMOTES MACROPHAGE-INDUCED IMMUNE TOLERANCE IN GLIOBLASTOMA THROUGH THE NMDAR/PPARΓ SIGNALING AXIS Article Open access 16 March 2023 MAIN The glioma

microenvironment orchestrates tumor evolution, progression and resistance to therapy1. In high-grade gliomas (HGG), microglia and monocyte-derived macrophages, collectively referred to as

glioma-associated myeloid cells (GAM), constitute up to 70% of the tumor mass2. Current concepts propose a recruitment of blood-borne macrophages to the glioma microenvironment, in which

phenotypic and functional shaping of invading macrophages and resident microglia is dependent on the tumor genotype, such as disease-defining mutations in the gene encoding IDH type 1 (IDH1)

that are causally related to profound tumor cell-intrinsic epigenetic and metabolic alterations associated with a favorable prognosis of patients with glioma2. Functionally altered GAM in

turn promote tumor growth by a variety of mechanisms. A striking feature of GAM is a poor antigen-presenting capacity and the acquisition of an immunosuppressive phenotype. While studies to

date suggest a continuum rather than a bimodal distribution of microglia-specific versus macrophage-specific genes in myeloid cells3, temporal cell type-specific functional states within the

glioma microenvironment have not been defined. Such analyses would reveal important molecular determinants of functional myeloid states as well as therapeutic targets within the myeloid

compartment. Here we comprehensively define longitudinal homeostatic and antigen-presenting myeloid cellular states, assess their tumor-genotype dependence and reveal underlying metabolic

mechanisms controlling them. RESULTS First, we performed RNA-seq on 30,000–600,000 sorted microglia and macrophages from 14 HGG (Fig. 1a and Supplementary Tables 1 and 2). Principal

component analysis showed a remarkable separation of samples based on the mutational status of IDH (Fig. 1a). To further dissect these genotype-dependent signatures, we performed single-cell

RNA-seq (scRNA-seq) on flow cytometry-purified CD45+CD3−CD19−CD20− hematopoietic cells isolated from IDH-wild-type (WT) (_n_ = 5) and IDH-mutant (_n_ = 5) HGG in comparison to control brain

tissues (_n_ = 7) (Fig. 1b). Seurat analysis of 4,460 cells that passed quality control using the data integration workflow revealed ten transcriptionally distinct clusters corresponding to

different cell types and states (Fig. 1c)4. Using hypergeometric tests for enrichments that considered different numbers of cells per condition, we found that myeloid cell clusters (C) C0,

C3, C4 and C6 were enriched in cells from control tissues, while C1 and C5 were enriched in cells from IDH-WT HGG, and C2 was enriched in cells from IDH-mutant HGG (Fig. 1d and Extended Data

Fig. 1a). Differential gene expression analysis of IDH-WT-enriched C1 and C5 showed a downregulation of microglial steady-state genes (_TMEM119_, _P2RY12_, _CSF1R_) with a concomitant

upregulation of interferon (IFN) signaling and hypoxia-associated genes, including _IFI44L_ and _HIF1A_, respectively. Furthermore, C1 and C5 expressed genes that are associated with acutely

activated macrophages, including _APOE_, _CD163_ and _S100A11_, suggesting that these clusters contain acutely infiltrating hematopoietic cells3,5. The IDH-mutant-enriched cluster C2 showed

upregulation of genes encoding chemokines (Fig. 1e). Control-enriched clusters C0, C3, C4 and C6 showed upregulation of microglia-defining genes such as _TMEM119_ and _OLFML3_ (Fig. 1e).

Accordingly, we found an increased cumulative expression of genes associated with homeostatic microglia (_P2RY12_, _CX3CR1_ and _CSF1R_) in the control-enriched clusters C0, C3, C4 and C6

(Fig. 1e). Differential gene expression analysis of cells assigned to microglia and macrophage clusters C0–C6 showed upregulation of major histocompatibility complex (MHC) class I and

II-coding genes, including _HLA-B_, _CD74_ and _HLA-DPA1_, in IDH-WT HGG-associated GAM with respect to their IDH-mutant HGG-associated counterparts (Fig. 1f and Extended Data Fig. 1b). The

latter cells displayed upregulation of steady-state microglia and inflammatory mediator-coding genes, such as _P2RY12_ and _IL1B_ (Fig. 1f). Next, we validated scRNA-seq findings at the

protein level using cytometry by time-of-flight (CyTOF). Unsupervised clustering of _n_ = 65,909 purinergic receptor (P2RY12)+ microglia cells (_n_ = 25,973 controls, _n_ = 17,646 cells from

IDH-mutant HGG and _n_ = 22,290 cells from IDH-WT HGG) identified ten clusters with similar protein expression profiles (Extended Data Fig. 1c–g). For ensuring robust clusterwise

comparisons of protein expression between conditions, we excluded four clusters (C4, C7, C8 and C10) containing less than 0.05% of all cells per condition (Fig. 1g). While clusters C1 and C5

consisted mostly of cells from control tissues, C3 predominantly contained cells derived from IDH-WT HGG, and C6 and C9 predominantly contained cells from IDH-WT and IDH-mutant HGG (Fig. 1g

and Extended Data Fig. 1e). C3 showed an activation gene expression profile with downregulation of homeostatic microglia proteins, including chemokine receptor CX3CR1, transmembrane protein

TMEM119 and P2RY12, as well as upregulation of microglial activation proteins, such as apolipoprotein (APOE) and receptor EMR1 (Fig. 1h and Extended Data Fig. 1g). Furthermore, clusters C6

and C9, which were enriched in both IDH-mutant and IDH-WT HGG, showed upregulation of antigen-presentation (AP)-associated proteins including HLA-DR and CD74 (Fig. 1h). Analysis of MHC class

II and co-stimulatory protein markers present in the antibody panel showed a consistent upregulation in HGG samples with respect to controls, with the exception of C2 (Fig. 1h). Notably,

myeloid cells were previously distinguished in immunogenic and tolerogenic states based on the expression of MHCII and co-stimulatory genes6. In line with transcriptomic profiling,

IDH-mutant HGG-derived myeloid cells showed a less pronounced downregulation of the microglia homeostatic signature and a less marked upregulation of the AP signature than IDH-WT HGG (Fig.

1h). Similar differences could be detected using a different CyTOF antibody panel (Supplementary Table 2 and Extended Data Fig. 2). Collectively, these data suggest a striking and

differential genotype-dependent shaping of GAM in human HGG toward an immunosuppressive phenotype. To investigate the dynamics and underlying molecular mechanisms of this glioma

genotype-dependent immunosuppressive phenotype of GAM, we made use of an experimental HGG mouse model, GL261 episomally overexpressing WT and mutant IDH (Fig. 2a and Extended Data Fig.

3a–g). We conducted scRNA-seq of flow cytometry-purified CD45+ cells isolated from IDH-mutant and IDH-WT GL261 gliomas comprising microglia, monocytes, macrophages, monocyte-derived

dendritic cells (DCs), mast cells, granulocytes and T and B cells at two time points during glioma progression: early after primary tumor establishment at day 7 (d7) and at a late-stage time

point at d28 after tumor injection (Fig. 2a–d). Expectedly, at d7, microglia made up to >75% of the myeloid cells in the tumor (75% ± 17.6% in IDH-WT samples and 92% ± 2.1% in IDH-mutant

samples), whereas, at d28, invading cells dominated the myeloid compartment (macrophages, 23% ± 11.6% in IDH-WT samples and 36% ± 7.1% in IDH-mutant samples; DCs, 11% ± 7.7% in IDH-WT

samples and 14% ± 3.2% in IDH-mutant samples; Fig. 2e). Interestingly, at d7, invading immune cells were more abundant in IDH-WT compared to IDH-mutant gliomas, while at the late stage,

hematopoietic immune cell contents were comparable between both experimental HGG (Fig. 2e). Comparative analysis considering cell type and IDH status of experimental HGG between d7 and d28

revealed a drop in relative microglia content accompanied by a concomitant increase in macrophage and DC numbers in IDH-mutant HGG, suggesting a higher influx of circulating immune cells

between early and late time points in these HGG. Importantly, these differences were not a result of differential tumor growth (Fig. 2e,f and Extended Data Fig. 3e). Based on these

observations, we hypothesized that, dependent on their IDH status, microglia shaped by the early HGG microenvironment drive differential recruitment of invading immune cells, particularly

blood-borne macrophages. Indeed, d7 microglia in IDH-WT tumors, in accordance with human IDH-WT HGG, show increased expression of _Ccl12_, the gene encoding the murine ligand of CCR2, which

is involved in peripheral myeloid cell recruitment to the central nervous system (CNS). _Ccl12_ was differentially upregulated in microglia from IDH-WT GL261 gliomas (Fig. 2g). Differential

expression analysis of microglia in early-stage experimental HGG further showed increased expression of genes encoding MHC and co-stimulatory molecules in microglia from IDH-WT gliomas,

while those from IDH-mutant gliomas displayed higher expression of steady-state microglia genes such as _P2ry12_ in accordance with our findings in human GAM (Fig. 2g). To assess gradual

changes between steady-state and activated microglia, we conducted a pseudotime analysis of early microglia using StemID2 (ref. 7) that showed a shift from a homeostatic to a differentiated

state. Consistent with our hypothesis, we found enrichment of steady-state microglia in IDH-mutant HGG and enrichment of activated microglia in IDH-WT HGG (Fig. 2h–j and Extended Data Fig.

3h,i). In summary, IDH-mutant experimental HGG showed an attenuated immunogenic microglia transcriptional profile and reduced content of infiltrating myeloid cells at the early time point

that increased during tumor progression. As late-stage experimental tumors were more enriched for recruited myeloid cells, we investigated functional phenotypes of monocyte-derived cells at

late time points and found robust expression of the previously defined AP signature in both cell compartments (Fig. 3a)8. While DCs showed only moderate differential signatures in IDH-mutant

compared to IDH-WT experimental tumors, macrophages from IDH-mutant gliomas showed upregulation of _Il1b_ with a concomitant downregulation of _Arg1_ (Fig. 3b,c). To assess mechanisms of

attenuating the AP signature in late-stage IDH-mutant HGG, we performed pseudotime analyses of macrophages and monocyte-derived DCs isolated from IDH-mutant compared to IDH-WT tumors (Fig.

3d–f and Extended Data Fig. 3h,i). Strikingly, we observed lower abundance of both cell types in IDH-mutant experimental HGG toward the end of each trajectory, suggesting either delayed

functional polarization or a differentiation block of monocyte-derived cells in late-stage IDH-mutant tumors (Fig. 3e,f). Comparing our murine and human datasets, the relatively high

intratumoral abundance of DCs was restricted to our HGG model, and current studies observe that DCs account for less than 5% of the human glioma immune cell infiltrate9. Therefore, we aimed

to functionally validate the attenuated AP signature inferred by transcriptomic profiling in microglia and macrophages. In ex vivo co-cultures of naive T cells and microglia and macrophages

isolated from experimental HGG at the late stage, we observed a consistent and ratio-dependent decrease in IFN-γ production and upregulation of programmed cell death protein (PD)-1 in both

cytotoxic and T helper cells. In addition, production of granzyme (Grz)B was reduced in cytotoxic T cells (Fig. 3g,h and Extended Data Fig. 3a). Strikingly, we observed a differential level

of T cell suppression by macrophages, but not by microglia, based on IDH-mutation status. Suppression of T cells was significantly increased in co-cultures with macrophages infiltrating

IDH-mutant experimental HGG. Based on previous observations in T cells10, we hypothesized that an attenuated AP signature of macrophages in IDH-mutant experimental tumors is dependent on the

neomorphic enzymatic activity of mutant IDH. Treatment of mice bearing intracranial IDH-mutant and IDH-WT HGG with the blood–brain-barrier-permeable mutant IDH inhibitor BAY 1436032 (ref.

11) revealed partial reversibility of an IDH-mutant-associated attenuated AP signature in macrophages but not in microglia (Fig. 3i). To define the molecular mechanism underlying this

time-dependent and tumor-genotype-dependent functionality shift, we exposed human monocytes and macrophages to the neomorphic enzymatic product of mutant IDH, _R_-2-hydroxyglutarate

(R-2-HG)12. Co-incubation of R-2-HG-pretreated monocytes or macrophages with T cells revealed a dose-dependent suppression of T cell proliferation (Fig. 4a). To validate that macrophage

exposure to R-2-HG leads to an attenuated AP capacity, we assessed CD80, CD86 and HLA-DR expression by flow cytometry. Indeed, a dose-dependent downregulation of these proteins after R-2-HG

exposure was observed (Fig. 4b,c). Macrophages and microglia took up exogenous R-2-HG independently of activation status (Fig. 4d and Extended Data Fig. 4a). Overexpression of amino acid

transporters known to transport R-2-HG, such as solute carrier (SLC)13A3, but not SLC22A6, SLC16A5 or SLC3A2, resulted in increased uptake of R-2-HG (Fig. 4e and Extended Data Fig. 4b)13. To

dissect the mechanism that mediates reprogramming of GAM in IDH-mutant tumors, we performed an in vitro transcriptome screen of R-2-HG-treated primary macrophages isolated from healthy

human donors (Fig. 4f and Extended Data Fig. 4c). Pathway analysis revealed that the top regulated pathway after R-2-HG exposure was induced by the synthetic toxin

2,3,7,8-tetrachlorodibenzo-_p_-dioxin (TCDD, Fig. 4f and Supplementary Table 3). TCDD is one of the strongest dioxin-like compounds to act via a specific ligand–receptor interaction with the

aryl hydrocarbon receptor (AHR)14. Strikingly, the expression of AHR target genes was significantly higher in the Cancer Genome Atlas (TCGA) datasets of IDH-mutant gliomas (_n_ = 226) than

in those of IDH-WT tumors (_n_ = 55) but only when normalized for intratumoral myeloid cell abundance assessed by _ITGAM_ (CD11b) expression (Extended Data Fig. 4d). AHR target genes, such

as _AHRR_ (encoding an AHR repressor), _TIPARP_ (encoding TCDD-inducible poly(ADP-ribose) polymerase) and _CYP1A1_ were induced in monocytes by R-2-HG to a similar extent as by the known

endogenous AHR ligand l-kynurenine (l-Kyn) in vitro (Fig. 4g). AHR target gene activation by R-2-HG in human macrophages was further observed to be dose dependent (Extended Data Fig. 4e). To

validate the specific expression of AHR target genes in GAM, we conducted pseudotime trajectory analyses of human glioma infiltrates. We found upregulation of AHR target genes along a

trajectory (C3 → C2) enriched in IDH-mutant-derived GAM but not IDH-WT-derived GAM (Fig. 4h). In contrast to this, IDH-WT-derived GAM were predominantly found along another, more immunogenic

trajectory with expression of antigen-presenting cell signature genes (C3 → C0 → C1, Fig. 4h). In accordance with the human dataset, all clusters forming the myeloid cell trajectory in

experimental HGG showed differential cumulative expression of an AHR activation gene signature between IDH-mutant-derived and IDH-WT-derived cells (Fig. 4i). As _AHR_ transcripts were most

abundant in monocytes as compared to other immune cells (Extended Data Fig. 4f), these findings are indicative of an immune cell subtype-specific vulnerability to reprogramming by R-2-HG.

Notably, AHR was identified as a critical cofactor for immunosuppressive transforming growth factor (TGF)-β and interleukin (IL)-1β signaling15,16. In addition, AHR directly promotes IL-10

production17,18. We found that, following exposure to R-2-HG, macrophages demonstrated an increased dose-dependent production of IL-10 and TGF-β in an AHR-dependent fashion (Fig. 4j and

Extended Data Fig. 4g). Analysis of monocyte-derived macrophages revealed that IL-10 production was in fact induced in an IDH-mutant experimental HGG microenvironment and required functional

AHR (Fig. 4k). Consequently, IL-10 levels in freshly isolated tumor lysates were higher in IDH-mutant experimental HGG when grown in an AHR-WT compared to an AHR-deficient microenvironment

(Fig. 4l). To verify whether AHR-dependent reprogramming of GAM by R-2-HG is a result of increased AHR translocation, we used an in vitro AHR translocation reporter assay in which GFP

expression is driven by a DRE-dependent promoter (_DRE_-GFP). Incubation with concentrated supernatants of glioma cell lines overexpressing mutant IDH resulted in increased AHR translocation

and transcriptional activity (Extended Data Fig. 4h). To determine whether the observed activation of AHR was driven by R-2-HG rather than by other differentially released substances, we

tested synthetic R-2-HG and observed increasing nuclear AHR translocation within 60 min of incubation (Fig. 4m and Extended Data Fig. 4i). We verified these results in an independent,

luciferase-based endpoint assay (Fig. 4n). Notably, R-2-HG-induced AHR translocation was comparable to that induced by l-Kyn (Fig. 4m–o and Extended Data Fig. 4i). As deprivation of l-Trp

reduced AHR translocation by R-2-HG, but not that by l-Kyn, our cumulative data suggested that R-2-HG is not a direct AHR ligand but leads to increased intracellular levels of l-Kyn,

presumably preceded by l-Trp catabolism and subsequent AHR activation (Fig. 4o). To assess whether R-2-HG-mediated suppression of T cells by macrophages is indeed dependent on l-Trp, we

performed co-culture assays with macrophages in l-Trp-free and control media, respectively. Strikingly, tryptophan deprivation of R-2-HG-exposed macrophages led to increased effector

functions of co-incubated T cells (Fig. 4p). Based on our findings, we hypothesized that immunosuppressive l-Trp catabolism via the kynurenine pathway drives reprogramming of macrophages

infiltrating IDH-mutant tumors19. To define dynamics of l-Trp metabolism in immune cells under the influence of R-2-HG, we undertook an LC–MS/MS-based study in macrophages and T cells (Fig.

5a,b and Extended Data Fig. 5a). Here we found that exogenous l-Trp was taken up by T cells in a dose-dependent fashion. While in T cells, a linear increase in l-Trp levels was accompanied

by a matching increase in l-Kyn levels; intracellular l-Trp levels remained stable in macrophages with increasing l-Kyn levels (Fig. 5a). Based on this observation, it is reasonable to

hypothesize that R-2-HG-exposed macrophages produce l-Kyn from imported l-Trp as a consequence of l-Trp catabolism via the kynurenine pathway. Tryptophan 2,3-dioxygenase (TDO2) and

indoleamine 2,3-dioxygenase (IDO)1 and IDO2 catalyze the rate-limiting step of the kynurenine pathway and together account for 90% of dietary l-Trp degradation20,21. The plasma

kynurenine-to-tryptophan ([l-Kyn][l-Trp]−1) concentration ratio has frequently been used to express or reflect the activity of these enzymes22. Indeed, when exposed to R-2-HG, macrophages

demonstrate a significant increase in the [l-Kyn][l-Trp]−1 ratio with increasing concentrations of extracellular l-Trp. By contrast, T cells do not show an increase in l-Kyn production

beyond a linear increase, which is the probable result of a shifted equilibrium reaction due to increased substrate levels (Fig. 5b). To identify the mechanism that underlies l-Trp

degradation in macrophages, but not in T cells, we performed cell-free enzymatic assays with described rate-limiting enzymes in the kynurenine pathway. Here, dose-dependent R-2-HG

supplementation resulted in increased kynurenine production by TDO2 but not by IDO1 or IDO2, suggesting a TDO2-inductive effect of R-2-HG (Fig. 5c–e). Based on induction kinetics, a new role

of R-2-HG as an allosteric activator within the TDO2 tetramer protein complex interface seems likely, as TDO2 is not dependent on α-ketoglutarate, similar to other enzymes that were

described to be affected by R-2-HG23,24. TDO2 is generally believed to be constitutively active in hepatocytes to achieve l-Trp homeostasis, while many cell types demonstrate low basal IDO

expression that can be rapidly induced by proinflammatory cytokines25. However, _TDO2_ expression analyses based on publicly available RNA-seq datasets across different immune cell

populations revealed a moderate expression level in monocytes and macrophages (Fig. 5f)26. In functional ex vivo assays using cells from _Tdo2_−/− mice, we demonstrated dependence of

R-2-HG-associated reduced T cell proliferation on _Tdo2_ expression in primary macrophages as well as IL-10 and TGF-β signaling. Interestingly, genetic ablation of _Tdo2_ in primary

macrophages reversed R-2-HG-associated reduced T cell proliferation when PD-ligand (L)1 immune checkpoint blockade was applied. Consequently, direct application of l-Kyn mimicked the

phenotypic effects of R-2-HG in TDO2-deficient cells (Fig. 5g). As _AHR_ and _TDO2_ transcripts are more abundant in human monocytes than in other immune cells (Extended Data Fig. 4f)27,

these findings are indicative of an immune cell subtype-specific vulnerability of macrophages to paracrine reprogramming by IDH-mutant tumors. We showed that, in macrophages, TDO2 is

directly induced by R-2-HG, leading to an accumulation of the AHR ligand l-Kyn. This pathological activation of the kynurenine pathway can only be sustained if decreasing l-Trp levels are

sensed and extracellular l-Trp is imported to a sufficient extent, as its de novo synthesis is impossible for animal cells. Interestingly, exposure of human monocyte-derived macrophages to

R-2-HG resulted in a similar amino acid transporter expression pattern as deprivation of extracellular l-Trp, suggesting that R-2-HG drives an amino acid starvation-like response as a result

of increased l-Trp degradation in macrophages (Fig. 6a). Remarkably, the expression of _SLC3A2_ was higher in R-2-HG-exposed cells than in l-Trp-deprived cells. Of note, not _Slc13a3_,

encoding the described transporter of R-2-HG, but _Slc3a2_ (CD98) contributed to the monocyte-to-macrophage trajectory in murine HGG (Fig. 3e). _Slc3a2_ and _Slc7a5_ (LAT1), forming the

heterodimer LAT1–CD98 upon translation, were differentially upregulated in IDH-mutant compared to IDH-WT experimental tumors and displayed consistent expression patterns over the

monocyte-to-macrophage trajectory (Fig. 6b,c). SLC3A2 (CD98) has multiple binding partners. Expression and in situ hybridization analyses of _Slc7a5_ across different tissue-resident

macrophage populations revealed highest levels of _Slc7a5_ in CNS macrophages (Fig. 6c and Extended Data Fig. 5b,c)28. It was shown that transmembrane transport of branched-chain amino acids

such as l-Trp is preferentially mediated by LAT1–CD9829 and might therefore provide the l-Trp needed to sustain activation of the kynurenine pathway by R-2-HG. Next, we investigated whether

increased LAT1–CD98-dependent l-Trp import attenuated the AP capacity of macrophages in experimental HGG via AHR signaling. To this end, we investigated LAT1–CD98 inhibition in vitro and in

vivo (Fig. 6d,e). Preconditioning of monocyte-derived macrophages with a small-molecule LAT1–CD98 inhibitor30 rescued the induction of 17 of 22 AHR targets that were induced by R-2-HG (Fig.

6d). Similarly, administration of the LAT1–CD98 inhibitor to glioma-bearing animals led to an increased abundance of MHCII+CD80+CD86+ immune-stimulatory macrophages in IDH-mutant tumors and

thus enhanced the AP signature (Fig. 6e). This is consistent with our previous finding that, in the context of macrophage exposure to R-2-HG, tryptophan deprivation led to increased

effector functions of co-incubated T cells (Fig. 4p). We aimed to investigate whether the microenvironment of human IDH-mutant gliomas is in fact configured for the maintenance of this

l-Trp-dependent axis. Using a matrix-assisted laser desorption–ionization (MALDI)-MS imaging (MSI)-based analysis approach, we discovered the extracellular accumulation of l-Trp in our human

HGG tissue cohort in situ (Fig. 6f and Extended Data Fig. 6a,b). As expected, all tumors diagnosed IDH mutant showed accumulation of R-2-HG, whereas there was no R-2-HG detectable in IDH-WT

HGG or control cortex samples (Fig. 6f,g). Interestingly, we observed a strong accumulation of extracellular l-Trp in IDH-mutant HGG across all human replicate sets that was significantly

higher than the abundance of l-Trp in IDH-WT HGG. l-Trp levels in IDH-WT HGG were moderate, while, in control cortex samples, they were comparable to the background level and thus

significantly lower than in tumor samples (Fig. 6f,g). Increased levels of l-Trp in tumor cell lines upon mutant IDH1 and/or IDH2 overexpression or exogenous R-2-HG exposure was previously

reported31. We similarly found a moderate-to-high increase in both intracellular and extracellular l-Trp levels, respectively, in IDH-mutant experimental HGG (Fig. 6h). While macrophages

showed moderate expression levels of _TDO2_ as compared to other immune cell subsets, RNA-seq analysis of TCGA glioma datasets showed no increase in _TDO2_ expression levels in IDH-mutant

tumors (Fig. 6i)32. Collectively, these data suggest that an overstimulated, R-2-HG-dependent uptake of l-Trp by myeloid cells via LAT1–CD98 maintains the immunosuppressive reprogramming of

GAM in IDH-mutant HGG. Our previous findings suggested that R-2-HG-induced T cell suppression by GAM is dependent on accumulation and degradation of l-Trp and functional AHR. Inhibition of

AHR or LAT1–CD98 by small-molecule inhibitors was effective in reverting R-2-HG-mediated reprogramming of macrophages as revealed by an AHR target screening array (Figs. 6d and 7a and

Extended Data Fig. 4h). When co-cultured with AHR-deficient macrophages exposed to R-2-HG, there was furthermore no consistent suppression of T cell proliferation or effector function,

irrespective of l-Trp abundance (Fig. 7b). We thus aimed to test whether IDH-mutant experimental HGG harbor a unique immunological vulnerability that could be therapeutically exploited by

AHR inhibition. We therefore combined T cell-activating immunotherapy by immune checkpoint blockade (anti-PD-L1 antibody) with a small-molecule AHR inhibitor33 to test the impact of the

AHR-mediated dysregulated immune microenvironment found in IDH-mutant HGG (Fig. 7c,d). In total, 40% of mice inoculated with IDH-WT HGG and treated with immune checkpoint blockade were

long-term survivors, whereas there were no responders to immune checkpoint blockade in mice bearing IDH-mutant tumors, supporting the observation that IDH-mutant tumors in fact foster a more

tolerogenic alignment of their immune microenvironment. While administration of an AHR inhibitor did not improve outcomes in mice bearing IDH-WT tumors (Fig. 7c), we observed statistically

significantly prolonged survival of mice bearing IDH-mutant tumors (Fig. 7d). In line with previous results (Figs. 3i and 4k), monocyte-derived macrophages from IDH-mutant gliomas treated

with the AHR inhibitor displayed a higher level of AP markers. Production of IL-10 and TGF-β by monocyte-derived cells was reduced by AHR inhibition in IDH-mutant tumors but not in IDH-WT

tumors (Fig. 7c,d). DISCUSSION Tumor-associated macrophages play a crucial role in a wide array of pathological hallmarks of tumors including gliomas2,34. In line with previous

studies35,36,37,38, our study defines biologically relevant functional states of GAM that are controlled by mutant IDH, a disease-defining driver mutation in gliomas39. Moving beyond

single-time-point assessments, we now longitudinally describe differential immune cell infiltration and phenotype dynamics during glioma progression that are orchestrated by a fluctuating

network of resident microglial cells and educated recruited myeloid cells. In late-stage experimental tumors, monocyte-derived macrophages further drive the tolerogenic alignment of the

glioma microenvironment. In IDH-mutant gliomas, we define the molecular mechanism as causally related to dynamic, R-2-HG-dependent tryptophan degradation by myeloid cells via TDO2 and

LAT1–CD98, resulting in the activation of the immunity master regulator AHR. Our study shows that R-2-HG not only profoundly shapes the glioma microenvironment but regulates targetable

immunosuppressive tryptophan catabolism in myeloid cells. The robustness of IDH genotype-dependent effects on the innate immune microenvironment in mouse and human datasets presented is

underscored by the recent recognition of considerable interspecies variation of microglial programs40. Our study not only supports the notion that HGG-associated microglia lose the

homeostatic gene expression signature present in normal brain tissue to enter a functionally altered cellular stage but also proposes a concept whereby the immunosuppressive phenotype in

IDH-mutant HGG is a result of an altered differentiation route of tumor-infiltrating monocytes. Importantly, and in addition to tumor cell-autonomous metabolic vulnerabilities conferred by

IDH mutations41,42,43, our study reveals an unexpected function of R-2-HG in regulating amino acid metabolism in immune cells. R-2-HG is taken up by myeloid cells to enzymatically induce

TDO2-dependent activation of the kynurenine pathway and, subsequently, the AHR. This pathological tryptophan degradation results in an amino acid starvation-like response that triggers the

expression of LAT1–CD98, a key transporter for tryptophan in proliferating cells25, which was previously linked to T cell activation and differentiation44,45. We here provide evidence that

LAT1–CD98 is critically involved in the differentiation and activation of GAM and that the previously observed altered amino acid metabolism in IDH-mutant gliomas31 is also responsible for

shaping an immunosuppressive tumor microenvironment through maintenance of this complex metabolic axis. We show that this regulatory metabolic network is particularly active in macrophages

infiltrating IDH-mutant gliomas through their distinct expression profile, which constitutes a metabolic vulnerability. Our study supports the hypothesis that gliomas with immunogenic driver

mutations, such as IDH1-R132H46, in addition to cell-autonomous epigenetic aberrations47, evolve by suppressing immune responses toward this neoantigen through production and secretion of

R-2-HG. Importantly, we show that, similar to tumor cell-autonomous metabolic vulnerabilities conferred by IDH mutations, specific R-2-HG-mediated immune vulnerabilities can be

therapeutically exploited to sensitize gliomas to immunotherapy. On a broader level, our data suggest that genetic alterations of driver genes, beyond tumor-intrinsic prognostic

implications, result in specific configurations of the immune microenvironment that can be therapeutically exploited by addressing tumor and immune cell metabolism. METHODS PROSPECTIVE

TISSUE COLLECTION AND HISTOPATHOLOGICAL VALIDATION Human control (_n_ = 7; mean age, 50 ± 19.07 years (21–74 years); sex, female:male (F:M)(4:3)), IDH1-WT HGG (_n_ = 5; mean age, 65.9 ± 14.9

years (32–81 years); sex, F:M(1)) and IDH1-R132H-mutant HGG (_n_ = 5; mean age, 48.5 ± 6.81 years (39–55 years); sex, F:M(0:5)) samples were prospectively collected from adult patients

undergoing brain surgery after informed consent. Patient sample collection at the Freiburg site was regulated under the ethics protocol 472/15. Briefly, after surgical removal, samples were

transferred into ice-cold PBS. A representative piece of the sample was snap frozen, cryosectioned, stained with H&E and examined by two board-certified neuropathologists. After

inclusion into the study, samples were fixed with formalin (4% paraformaldehyde in PBS) and embedded in paraffin for in situ validation. The remaining sample was maintained on ice at all

times. After removal of the meninges, tissue samples were transferred into ice-cold HBSS containing 10 mM glucose and 10 mM HEPES and mechanically dissociated using glass shearing with a

10-ml Potter-Elvehjem pestle and glass-tube homogenizer (Merck). The suspension was passed through a 70-µm cell strainer (BD Biosciences). Myelin was removed using gradient centrifugation

with 37% Percoll (Merck) at 800_g_ for 30 min without a brake. Cell pellets were dissolved and cryopreserved in FCS:DMSO (9:1; Merck). FLOW CYTOMETRY SORTING HUMAN SAMPLES A MoFlo Astrios

(Beckman Coulter) was used for cell sorting. The cell suspension was stained with the following antibodies: anti-CD45 (clone HI30, BD Biosciences), anti-CD11b (clone M1/70, eBioscience),

anti-CD3 (clone SP34–2, BD Biosciences), anti-CD19 (clone SJ25C1, BioLegend) and anti-CD20 (clone 2H7, BioLegend). Before surface staining, Fc receptors were blocked (BD Biosciences).

DAPI-positive cells were excluded (1:1,000). MURINE SAMPLES Brain tumors were digested with Liberase D (50 µg ml−1) and meshed through a cell strainer to obtain a single-cell suspension.

Myelin was removed using a continuous 30% Percoll gradient. The cell suspension was stained with the following antibodies after blocking Fc receptors (anti-mouse CD16/CD32 antibody,

BioLegend): anti-CD45.2 (clone 30-F11, BioLegend), anti-CD11b (clone M1/70, BioLegend), anti-NK1.1 (clone REA1162, Miltenyi Biotec), anti-CD3 (clone 17A2, BioLegend) and fixable viability

dye eFluor 780 (Invitrogen). Cells were sorted under sterile conditions on a BD FACSAria Fusion equipped with the following lasers: 405 nm, 488 nm, 561 nm and 640 nm, using an 85-µm nozzle

and four-way purity mode. SINGLE-CELL RNA SEQUENCING, TRANSCRIPT QUANTIFICATION AND ANALYSIS Single-cell transcript amplification and library preparation were performed using mCEL-seq2 as

previously described48,49. Paired-end reads were aligned using BWA (version 0.6.2-r126) with default parameters50 to a transcriptome containing all gene models based on human ENCODE release

version 24. Isoforms of a given gene were treated as one gene locus. The right mate of each read was mapped to an ensemble of all gene loci and 92 ERCC spike-ins in the sense direction51.

The left mate contained the unique molecular identifier (UMI, six bases) and a cell-specific barcode (six bases), followed by a poly-T stretch. The number of distinct UMIs was recorded for

each gene locus and assigned to each cell based on the cell barcode. Transcript counts were aggregated from UMI aggregate counts based on binomial statistics52. HUMAN SINGLE-CELL RNA

SEQUENCING DATA ANALYSIS Samples were sequenced in batches of eight libraries with 192 cells each per lane on an Illumina HiSeq 3000 sequencing system (paired-end multiplexing run) at a

depth of ~130,000–200,000 reads per cell. Fifty-two libraries with 4,460 cells after quality control (control, 2,494; IDH1-WT HGG, 1,000; IDH1-mutant HGG, 966) were integrated and analyzed

using Seurat version 3 (ref. 53). Cells expressing >2% of _KCNQ1OT1_, a previously identified marker of cells with low quality7, were excluded from analysis. Also, transcripts with a

correlation coefficient >0.65 with _KCNQ1OT1_ were excluded. A total of 32,088 gene loci were quantified with a median of 1,296 ± 753.55 (s.d.) genes per cell. Data analysis was performed

using the Seurat version 3 pipeline. The counts table was filtered for features expressed by at least three cells and cells with at least 500 detected features, corresponding to the

arguments ‘min.cells=3’ and ‘min.features=500’ in the CreateSeuratObject function call. Data were scaled and normalized using the SCTranform function53 with the function set to return 10,000

variable features and regress out the percentage of mitochondrial genes, corresponding to the arguments ‘variable.features.n=10,000’, ‘return.only.var.genes=F’ and

‘vars.to.regress=”percent.mt”’. For dataset integration, 10,000 variable features were used with batch-effect associated features that contained the following patterns in their name filtered

out: _JUN_, _FOS_, _RP_, _ZFP36_, _EGR_, _HSP_, _MALAT1_, _XIST_, _MT_- and _HIST_. These genes were subsequently also removed from the RunPCA function. Subsequently, UMAP and shared

nearest-neighbors graph construction were performed on the top 15 principal components. Clusters were identified with resolution set to 0.6. ANALYSIS OF THE MOUSE 10X SINGLE-CELL RNA

SEQUENCING DATA First, 10x libraries were prepared from CD45+ cells purified by flow cytometry in a single run. Read alignment and transcript quantification were conducted using Cell Ranger

version 3.1.0. The resulting counts files were analyzed using the Seurat version 3 dataset integration workflow. The counts table was filtered for features expressed by at least five cells

and cells with at least 500 detected features, corresponding to the arguments ‘min.cells=5’ and ‘min.features=500’ in the CreateSeuratObject function call. Data were scaled and normalized

using the SCTranform function, with the function set to return 10,000 variable features and regress out the percentage of mitochondrial genes, corresponding to the arguments

‘variable.features.n=10,000’, ‘return.only.var.genes=F’ and ‘vars.to.regress=”percent.mt”’. For dataset integration, 10,000 variable features were used with batch-effect associated features

that contained the following patterns in their name filtered out: _Jun_, _Fos_, _Gm_, _Rpl_, _Rps_, _Atf3_, _Zfp36_, _AY_, _Egr_, _Hsp_, _Malat1_, _Xist_, _mt_-, _Hist_ and _Socs3_. These

genes were subsequently also removed from the RunPCA function. Subsequently, UMAP and shared nearest-neighbors graph construction were performed on the top 15 principal components. Clusters

were identified with resolution set to 2.5. DIFFERENTIAL GENE EXPRESSION ANALYSIS OF SINGLE-CELL RNA SEQUENCING DATA Differential gene expression analysis was conducted using the

FindAllMarkers function of Seurat version 3. For comparisons between two conditions, the FindMarkers function was used. Features with an average log fold change greater than 0.25 and an

adjusted _P_ value less than 0.05 were considered significant. CLUSTER ENRICHMENT ANALYSIS Enrichment analysis for a given condition in a cluster was conducted using a hypergeometric test

implemented in R under the phyper function. This test considers the number of cells from condition _x_ in a given cluster with respect to all cells from condition _x_ in the dataset, all

cells from condition _y_ in the dataset and the number of cells in a given cluster. We used it to calculate the probability that number _n_ or more cells from condition _x_ could be found in

a given cluster by chance. Statistical significance was assumed for probabilities <0.05. Correction for multiple testing was achieved using the Benjamini–Hochberg method. CLUSTERWISE

COMPARISON OF GENE EXPRESSION To compare gene expression between clusters and conditions, we fit a generalized linear model with a negative binomial link function using the ‘glm.nb’ function

of the MASS R package. Pairwise testing was achieved by calculating the estimated marginal means for comparisons of diagnoses per cluster using the emmeans function of the emmeans R package

with Tukey adjustment. PSEUDOTIME ANALYSIS OF SINGLE-CELL TRANSCRIPTOMES Pseudotime trajectory analysis of scRNA-seq data was conducted using the StemID2 functionality of RaceID3 and the

FateID R package with default settings7,48. First, a lineage tree was computed using the nearest-neighbor mode (nmode=TRUE) with default parameters of StemID2. Next, a list of significant

links determined in the previous step was chosen based on the underlying question. A filtered gene expression matrix was obtained through the getfdata function of the FateID R package and

used as input for pseudotime gene expression analysis of cells along the given list of links. Genes expressed at less than two normalized transcripts in at least ten cells in mice were

filtered out using the filterset FateID function. Genes with similar gene expression profiles were grouped into modules on a self-organizing map using the getsom FateID function with the

minimal size of modules set to 3 and the correlation threshold set to 0.85. With the help of the procsom function, modules on self-organizing maps were grouped into larger modules that were

used for visualization and downstream gene ontology term and other analyses. BULK RNA SEQUENCING OF GL261 IDH1-WT AND IDH1-R132H CELL LINES Bulk RNA-seq data from GL261 IDH1-WT and

IDH1-R132H cell lines were analyzed on the usegalaxy.eu platform. Raw FastQ files were mapped to the mm10 genome using STAR aligner version 2.7.5b followed by featureCounts version 1.6.4.

For figures, RPKM (reads per million kb) values were compared. CYTOF SAMPLE PREPARATION AND MEASUREMENT CyTOF was conducted as previously described54 using intracellular barcoding for mass

cytometry. Briefly, cells pelleted by Percoll-gradient centrifugation were fixed with a fixation–stabilization buffer and frozen at –80 °C until analysis. Thawed cells were barcoded using

premade combinations of six different palladium isotopes: 102Pd, 104Pd, 105Pd, 106Pd, 108Pd and 110Pd (Cell-ID 20-Plex Pd Barcoding kit, Fluidigm) and pooled for further processing. The

resulting cell pellet was resuspended in 100 µl antibody cocktail specific for surface markers (Supplementary Tables 1 and 2). Of note, we excluded 169Tm-TGF-β (antibody panel B) from the

analysis because of high background in this channel. For intracellular staining, stained cells were subsequently incubated in fixation–permeabilization buffer (Fix/Perm Buffer, eBioscience)

for 60 min at 4 °C, washed with permeabilization buffer (eBioscience) and stained with antibody cocktails against intracellular molecules in permeabilization buffer for 1 h at 4 °C. Cells

were subsequently washed twice with permeabilization buffer and incubated overnight in 2% methanol-free formaldehyde. The next day, cells were washed and resuspended in iridium intercalator

solution (Fluidigm) for 1 h at room temperature. Afterward, samples were washed with cell-staining buffer and ddH2O (Fluidigm). Cells were pelleted and kept at 4 °C until CyTOF measurement.

Cells were analyzed using a CyTOF2, upgraded to Helios specifications, with software version 6.5.236. Instrument and acquisition settings were set up as described previously54. Mass

cytometry data processing and analysis were performed as described previously using Cytobank and CATALYST54. Clusterwise comparisons of protein expression between conditions was conducted on

clusters that contained at least 0.05% of all cells per condition to ensure robust comparisons (>31 cells for antibody panel A and >22 cells for antibody panel B). Clusters that were

below this threshold were not considered for visualization or analysis. Pairwise testing was performed using the one-way Kruskal–Wallis test followed by Dunn’s post hoc test with

multiple-testing adjustment according to the Benjamini–Hochberg method. PERIPHERAL BLOOD MONONUCLEAR CELLS Peripheral blood mononuclear cells were isolated from research-only buffy coat

formulations from healthy donors or patients from the Neurology Clinic Heidelberg upon patient consent. EDTA was used as an anticoagulant. Blood formulations were kept at 4 °C before further

processing. EXPERIMENTAL ANIMALS C57BL/6J WT mice were purchased from Charles River. B6.Tdo2tm1Tnak (_Tdo2_–/–) and B6;129-Ahrtm1Bra/J (_Ahr_–/–) mice were bred according to local

regulatory authorities (breeding approval reference EP-Z124I02). All mice were 7–10 weeks of age at use. Mice were kept under specific pathogen-free conditions at the animal facility in the

DKFZ Heidelberg. TUMOR CELL INOCULATION GL261 tumor cells (1 × 104 cells) were diluted in 2 µl sterile PBS (Sigma-Aldrich) and stereotactically implanted into the right hemisphere of

7–10-week-old female C57BL/6J mice (coordinates, 2 mm right lateral to the bregma and 1 mm anterior to the coronal suture with an injection depth of 3 mm below the dural surface) using a

10-µl Hamilton microsyringe driven by a fine-step stereotactic device (Stoelting). CELL LINES The murine glioma cell line GL261 was obtained from the Division of Cancer Treatment and

Diagnosis at the National Cancer Institute. GL261 cells were cultured in DMEM (Sigma-Aldrich), supplemented with 10% FBS (Sigma-Aldrich), 100 U ml−1 penicillin and 100 μg ml−1 streptomycin

(Invitrogen). The embryonic kidney cell line HEK293 was obtained from ATCC and sold by LGC Standards. This cell line was cultured in DMEM (Sigma-Aldrich), supplemented with 10% FBS

(Sigma-Aldrich), 100 U ml−1 penicillin and 100 μg ml−1 streptomycin (Invitrogen). DETERMINATION OF TRYPTOPHAN METABOLITES Frozen cell pellets were processed following an adjusted extraction

protocol targeting tryptophan and kynurenine metabolites55,56. Briefly, samples were disrupted in 100 µl acidified mobile phase (0.2% formic acid with 1% acetonitrile in water) and 400 µl

ice-cold methanol using a sonication bath (Transsonic 460, Elma) for 15 min at the highest frequency. Afterward, samples were kept at −20 °C for 15 min to precipitate all protein.

Subsequently, samples were centrifuged for 15 min at 4 °C and 16,400_g_, and the resulting supernatant was transferred to a new 1.5-ml microcentrifuge tube (Eppendorf). Finally, the

supernatant was dried using the Eppendorf Concentrator Plus, set to no heat, and resuspended in 40 µl acidified mobile phase. For metabolite separation and detection, an ACQUITY I-class PLUS

UPLC system (Waters) coupled to a QTRAP 6500+ (SCIEX) mass spectrometer with an electrospray ionization source was used. In detail, metabolites were separated by reversed-phase

chromatography on an ACQUITY HSS T3 column (150 mm × 2.1 mm, 1.7 µm, Waters) kept at 20 °C with a flow rate of 0.4 ml min−1. An overview of multiple-reaction monitoring transitions that were

used can be found in Supplementary Table 5. Clear separation of tryptophan and tryptophan-derived compounds was achieved by increasing the concentration of solvent B (acetonitrile with 0.1%

formic acid) in solvent A (water with 0.1% formic acid) as follows: 1 min, 5% B; 11 min, 40% B; 13 min, 95% B; 15 min, 95% B; and return to 5% B in 5 min. Data acquisition and processing

was performed with the SCIEX OS software suite (SCIEX). RT–QPCR For isolation of RNA, cells were lysed without prior washing using TRIzol reagent (Thermo Fisher Scientific) and purified

using the RNeasy MinElute Cleanup kit (Qiagen). Next, 1 μg total mRNA was used for cDNA synthesis using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems) according to the

manufacturer’s instructions. cDNA was synthesized as described above. RT–qPCR was performed using the primaQuant qPCR SYBR Green Master Mix with ROX (Steinbrenner), and samples were run on

a QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific). All samples were analyzed in quadruplicate, and melting curves were considered to evaluate PCR reactions. Ct values were

normalized to both _GAPDH_ and _RPL9_ (human), or _Rpl13_ (murine), respectively. Primer-only reactions served as negative controls. All primers were checked for primer efficiency by RT–qPCR

and serial dilution of cDNA and were used if efficiency was >90%. TREATMENT WITH _R_-2-HYDROXYGLUTARATE IN VITRO d-α-hydroxyglutaric acid disodium salt (≥95.0% purity, determined by gas

chromatography) was obtained from Sigma-Aldrich and reconstituted in PBS (Sigma-Aldrich) at a concentration of 2 M. Cells were treated with R-2-HG by diluting the 2 M stock solution in the

respective cell medium. R-2-HG-containing medium was then pulse vortexed and added to the cells. [3H]THYMIDINE PROLIFERATION MEASUREMENTS If not mentioned otherwise, T cells or mixed cells

from co-cultures were seeded at 500,000 cells per well in a 96-well plate as technical quadruplicates and pulsed with RPMI 1640 (PAN-Biotech) with 10% FBS and human serum AB (both from

Sigma-Aldrich), 100 U ml−1 penicillin and 100 μg ml−1 streptomycin (Invitrogen), 2 mM l-glutamine (Invitrogen) and 50 μM β-mercaptoethanol (Sigma-Aldrich), supplemented with

[methyl-3H]thymidine (PerkinElmer), resulting in a radioactivity concentration of 20 mCi ml−1. Incorporation of radioactively labeled thymidine was allowed for 18 h, after which cells were

shock frozen and kept at −20 °C. Scintillation counting was performed to determine radionuclide uptake using a cell harvester (Tomtec) and a scintillation counting device (Wallac MicroBeta

TriLux Scintillation Counter, PerkinElmer). Proliferation measurements were given in counts per minute. CYTOKINE ELISA Primary human macrophages and dendritic cells were treated as described

above and incubated for 72 h after unspecific stimulation with LPS (Sigma-Aldrich) and recombinant human IFN-γ (PeproTech). Supernatants were transferred to cytokine-specific

antibody-coated ultra-low-binding 96-well plates (Corning), and cytokine ELISAs were performed using horseradish peroxidase-conjugated antibodies according to the manufacturer’s instructions

(eBioscience). The development process was stopped with 1 M H2SO4, and optical density (OD) was measured at 570 nm and 450 nm. Cytokine concentrations were calculated based on OD450 nm−570

nm according to parallel serial dilutions of cytokine standards included in the respective ELISA kits. ELISA detection was used for the following human cytokines: IL-10, TGF-β (ELISA

Ready-SET-Go! kits, eBioscience). Murine cytokines were measured accordingly. AHR ACTIVITY ASSAYS Assays were performed as described previously57. Briefly, HEK293T cells were transfected in

96-well plates with a combination of plasmids for expressing AHR and ARNT and reporter plasmids: either the dioxin-responsive element (_DRE_)-GFP reporter (Qiagen) or the pGudLuc7.1F Cignal

xenobiotic response element (_XRE_) reporter. Cells were transfected using FuGENE HD (Promega) following the manufacturer’s instructions (FuGENE HD:DNA ratio of 3:1). Cells were cultured

with the indicated compounds or GL261 cell line supernatant, and the activity of the reporter was measured after 6 h using a Promega luciferase assay kit (E1500) following the manufacturer’s

instructions. Luminescence was measured using a PHERAstar FS plate reader (BMG LABTECH). Data represent the mean of three cell lines normalized to the highest intensity result, where,

within each cell line replicate, each condition was repeated in triplicate. MALDI FOURIER-TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY IMAGING For MALDI-MSI analyses, frozen tissue

sections on ITO slides were dried in a vacuum for 15 min at room temperature and subsequently spray-coated with 1,5-DAN prepared at 10 mg ml−1 with a 50% acetonitrile (vol/vol) matrix using

an HTX TM-Sprayer (HTX Technologies). The matrix deposition protocol consisted of ten layers sprayed at a matrix flow rate of 100 µl min−1 and a spray-head velocity of 1,200 mm min–1 with a

distance of 3 mm between sprayed lines (HH pattern). The spray nozzle height was set to 40 mm from the ITO slide, and temperature was increased to 60 °C with a pressure of 10 psi and a gas

flow rate of 2 l min−1. High-resolution data acquisition was performed using the 7T Fourier-transform ion cyclotron resonance (FTICR) mass spectrometer (SolariX XR, Bruker Daltonics)

equipped with an Apollo II dual MALDI/ESI ion source and a 2-kHz Smartbeam II laser. MSI data were recorded in negative-ion mode within a _m_/_z_ range of 100–12,000 using a raster width of

50 μm and 100 laser shots per pixel at a laser power of 18%. Spectra were recorded using 1 million transient data points (FID 0.4893 s) with an online calibration using an internal lock mass

of _m_/_z_ 157.076025 (deprotonated 1,5-DAN peak). The following parameters were used: ion transfer (funnel 1, 150 V; skimmer 1, 15 V; funnel RF amplitude, 150 Vpp), octopole (frequency, 5 

MHz; RF amplitude, 350 Vpp), collision cell (RF frequency, 2 MHz; RF amplitude, 1,900 Vpp), transfer optics (time of flight, 0.5 ms; frequency, 6 MHz; FR amplitude, 350 Vpp); quadrupole (Q1

mass, 140 _m_/_z_); excitation mode (sweep excitation, sweep step time, 15 μs) and data reduction and storage (profile spectrum was saved with a data reduction factor of 97%). Data

acquisition was performed using ftmsControl 2.2.0 from Bruker Daltonics, and measurement regions were specified using flexImaging 5.0x64 (Bruker Daltonics). MSI data were acquired from each

tissue section as well as from matrix control areas to check for matrix interference on analytes. A mixture of the following amino acids and compounds was used as an external quadratic

calibration for the MALDI-FTICR-MS instrument: the deprotonated 1,5-DAN peak (_m_/_z_ 157.076025), kynurenic acid (_m_/_z_ 188.034219), tryptophan (_m_/_z_ 203.081504), kynurenine (_m_/_z_

207.076419), sunitinib (_m_/_z_ 397.203431), olanzapine (_m_/_z_ 311.132494), sorafinib (_m_/_z_ 463.077929) and CZC54252 (_m_/_z_ 503.126278). FLOW CYTOMETRY For intracellular cytokine

staining, cells were incubated with 5 μg ml−1 brefeldin A (Sigma-Aldrich) for 5 h at 37 °C with 5% CO2 to allow for intracellular enrichment of cytokines. Brain tumor and spleen cell

suspensions were blocked with anti-CD16/CD32 (eBioscience, 93, 14-0161), and extracellular targets were stained for 30 min at 4 °C (Supplementary Table 4). For detection of intracellular

antigens, cells were fixed, permeabilized and stained using the FOXP3 Transcription Factor Staining buffer set (eBioscience, 00-5523) and the antibodies listed in Supplementary Table 4.

Staining of intracellular targets was performed for 45 min at 4 °C. Stained lymphocytes were analyzed on the FACSCanto II or the LSRFortessa (BD Biosciences) or on the Attune NxT (Thermo

Fisher). BD FACSDiva software version 9 and FlowJo version 9 or 10 were used for data analysis. STATISTICS AND REPRODUCIBILITY SINGLE-CELL PROFILING For Figs. 1–3 and Supplementary Fig. 1,

no statistical method was used to predetermine sample size. Experiments were not randomized. Investigators were not blinded to allocation during experiments or outcome assessment. For Fig.

1d, hypergeometric testing was used to test for enrichment. Genes expressed by less than three cells and cells with less than 500 detected genes were excluded from the analyses. For Fig. 1e,

differentially expressed genes were determined based on the Wilcoxon rank-sum test using the FindAllMarkers Seurat function. The expression of the top 15 marker genes per cluster is shown.

Genes expressed by less than three cells and cells with less than 500 detected genes were excluded from the analyses. Furthermore, mitochondrial genes and genes associated with the

dissection response were excluded. The pattern these gene names contained are the following: _HTRA_, _LIN_, _EEF_, _CTC_-, _MIR_, _CTD_-, _AC0_, _RP_, _FOS_, _JUN_, _MTRNR_, _MT_-, _XIST_,

_DUSP_, _ZFP36_, _RGS_, _PMAIP1_, _HSP_, _NEAT1_, _HIST_ and _MALAT1_. For Fig. 1f, genes differentially expressed between the two indicated clusters were determined based on the Wilcoxon

rank-sum test using the FindMarkers Seurat function. Up to the five top marker genes per cluster are indicated. Genes expressed by less than three cells and cells with less than 500 detected

genes were excluded from the analyses. For Fig. 1h, a Kruskal–Wallis test was conducted followed by Dunn’s test for multiple comparisons. Multiple-testing adjustment was performed using

Holm’s method. Clusters with less than 1‰ cells per condition were excluded from the analyses. For Fig. 2e, hypergeometric testing was used to test for enrichment. Genes expressed by less

than ten cells and cells with less than 500 detected genes were excluded from the analyses. For Fig. 2f, mean cell type percentages within each glioma genotype were compared based on the

negative binomial function. Pairwise comparisons were performed by calculating the estimated marginal means for comparisons of diagnoses per cluster with Tukey adjustment. Genes expressed by

less than ten cells, cells with less than 500 detected genes and cells with absolute average counts less than 20 were excluded from the analysis. For Fig. 2g, genes differentially expressed

between microglia from the two glioma genotypes were determined based on the Wilcoxon rank-sum test using the FindMarkers Seurat function. Up to the five top marker genes per cluster are

indicated. Genes expressed by less than ten cells and cells with less than 500 detected genes were excluded from the analyses. For Fig. 2h, statistically significant intercluster links were

determined using the StemID2 algorithm. Genes expressed by less than ten cells with less than two detected transcripts were excluded from the analyses. For Fig. 2i, differentially expressed

genes in the clusters along the trajectory from Fig. 2h were determined based on the Wilcoxon rank-sum test using the FindAllMarkers Seurat function. The expression of the top 15 marker

genes per cluster is shown. Genes expressed by less than ten cells and cells with less than 500 detected genes were excluded from the analyses. For Fig. 2j, gradual gene changes along the

indicated cell trajectory were determined using the StemID2 algorithm. Genes with a correlation >0.85 were arranged into modules with similar expression patterns. Genes expressed by less

than ten cells with less than two detected transcripts and modules with less than three genes were excluded from the analyses. The Spearman correlation in the bottom panel was calculated in

the bottom panel. For Fig. 3b,c, genes differentially expressed between the indicated clusters were determined based on the Wilcoxon rank-sum test using the FindMarkers Seurat function.

Representative top differentially expressed genes per cluster are indicated. Genes expressed by less than ten cells and cells with less than 500 detected genes were excluded from the

analyses. For Fig. 3d, statistically significant intercluster links were determined using the StemID2 algorithm. Genes expressed by less than ten cells with less than two detected

transcripts were excluded from the analyses. For Fig. 3e,f, gradual gene changes along the indicated cell trajectory were determined using the StemID2 algorithm. Genes with a correlation

>0.85 were arranged into modules with similar expression patterns. Genes expressed by less than five cells with less than two detected transcripts and modules with less than three genes

were excluded from the analyses. The Spearman correlation in the bottom panel was calculated in the bottom panel. For Fig. 4i, left, the analysis corresponds to that in Fig. 3d. For Fig. 4i,

right, statistical significance of the interglioma genotype for the respective clusters was determined by fitting a generalized linear model with a negative binomial link function. Pairwise

comparisons were performed by calculating the estimated marginal means for comparisons of diagnoses per cluster with Tukey adjustment. Genes from the indicated expression signature were

included. For Supplementary Fig. 1b, hypergeometric testing was used to test for enrichment. No statistical method was used to predetermine sample size. Genes expressed by less than three

cells and cells with less than 500 detected genes were excluded from the analyses. For Supplementary Fig. 1c, the statistical significance of each diagnosis for the respective clusters was

determined by fitting a generalized linear model with a negative binomial link function. Pairwise comparisons were performed by calculating the estimated marginal means for comparisons of

diagnoses per cluster with Tukey adjustment. Genes from the indicated expression signature were included. For Supplementary Fig. 1h,j, a Kruskal–Wallis test was conducted, followed by Dunn’s

test for multiple comparisons. Multiple-testing adjustment was performed using Holm’s method. Clusters with less than one promille cells per condition were excluded from the analyses. For

Supplementary Fig. 2h,i, the indicated lineage graphs were computed using StemID2 in neighbor mode with a _P_ value cutoff of 0.01, and the scthresh parameter for links shown in the graph

was set to 0.9 IN VITRO AND IN VIVO EXPERIMENTS For Figs. 2–7 and Extended Data Figs. 2–4, no statistical method was used to predetermine sample size. Experiments were not randomized.

Investigators were not blinded to allocation during experiments or outcome assessment. For in vitro experiments, unless stated otherwise in figure legends, data are represented as individual

values or as mean ± s.e.m. Group sizes (_n_) and applied statistical tests are indicated in figure legends. Significance was assessed by either unpaired _t_-test analysis, paired _t_-test

analysis or one-way ANOVA analysis with Tukey post hoc testing as indicated in figure legends. Spearman correlation was applied for all correlation analyses, and the Kaplan–Meier method was

used to examine survival differences. Statistics were calculated using GraphPad Prism 7.0. Key experiments (TAM phenotyping, AHR reporter assays) were all performed at least three times or

with biologically independent healthy human donors or mice. All other experiments were performed as specified in figure legends. For in vivo experiments, unless stated otherwise in figure

legends, data are represented as individual values or as mean ± s.e.m. Group sizes (_n_) and applied statistical tests are indicated in figure legends. Significance was assessed by either

unpaired _t_-test analysis, paired _t_-test analysis or one-way ANOVA analysis with Tukey post hoc testing as indicated in figure legends. Survival was analyzed by the log-rank Mantel–Cox

test. Sample size was calculated with the help of a biostatistician using R version 3.4.0. Assumptions for power analysis were as follows: _α_ error, 5%; _β_ error, 20%. Values for standard

deviations and differences between experimental groups were based on previous experiments (whenever a similar data type was available). In all other cases, a pilot group size was used

without using a statistical method to predetermine sample size. Mice were randomized into treatment groups stratified for tumor size (measured by MRI) at the time when treatment started.

Intracranial tumor experiments were performed in a blinded manner (MRI, treatment, flow cytometric analyses). In case animals had to be killed before the pre-defined endpoint (due to weight

loss or other termination criteria), they were excluded from any downstream analyses. All investigators were blinded to allocation during experiments and outcome assessment. MALDI

FOURIER-TRANSFORM ION CYCLOTRON RESONANCE MASS SPECTROMETRY IMAGING After being acquired, all MALDI-FTICR-MSI datasets were imported directly into R 3.6.0 (R Foundation for Statistical

Computing) and processed using the MALDIquant package58. Datasets were acquired with on-the-fly centroid detection natively run on the measurement equipment, and, therefore, the imported

datasets represented already centroided mass spectra peak-picked with a signal-to-noise ratio of 3. Resulting mass spectra were normalized to their total ion count. All subsequent analyses

and visualization were performed in R. Mass resolution, which was calculated based on the full width at half maximum of analytes’ peaks, was approximately 200,000 for R-2-HG and 145,000 for

both l-Trp and l-Kyn. After acquisition, the mass accuracy for the amino acid mixture mentioned in the previous section was recalculated on a pixel-wise basis. The maximum mass shift

observed was <2.5 ppm with a median mass shift <1 ppm for all amino acid mixture analytes. To avoid picking spurious signals originating from tissue-cutting artifacts and tears,

background matrix pixels were computationally dropped based on their total ion count and corresponding optical image pixel gray level intensity. To estimate the level of the detected

intensities of the analytes of interest, a search window of 2 ppm (of the theoretical mass) was used to search for the presence of the corresponding analyte. The corresponding signal

intensity was then collected and rescaled to an intensity range of (0, 1) to simplify visualization. REPORTING SUMMARY Further information on research design is available in the Nature

Research Reporting Summary linked to this article. DATA AVAILABILITY Bulk and scRNA-seq data that support the findings of this study were deposited in the Gene Expression Omnibus under the

SuperSeries accession code GSE166420. This consists of the following SubSeries: GSE166218 (mouse 10x data), GSE166418 (human CEL-seq2 data) and GSE166521 (GL261 bulk RNA-seq data). Mass

cytometry data were deposited in the FlowRepository at https://flowrepository.org/id/FR-FCM-Z3G7. TCGA dataset was downloaded from http://gliovis.bioinfo.cnio.es/. Imaging source data for

this manuscript can be found at https://doi.org/10.6084/m9.figshare.14166983. All other data supporting the findings of this study are available from the corresponding author on reasonable

request ([email protected]). Source data are provided with this paper. CODE AVAILABILITY Custom code for transcriptomic and proteomic analyses can be found at

https://github.com/rsankowski/friedrich-et-al-IDH1wt-mut-micr.git. REFERENCES * Broekman, M. L. et al. Multidimensional communication in the microenvirons of glioblastoma. _Nat. Rev.

Neurol_. 14, 482–495 (2018). * Gutmann, D. H. & Kettenmann, H. Microglia/brain macrophages as central drivers of brain tumor pathobiology. _Neuron_ 104, 442–449 (2019). Article  CAS 

Google Scholar  * Venteicher, A. S. et al. Decoupling genetics, lineages, and microenvironment in IDH-mutant gliomas by single-cell RNA-seq. _Science_ 355, eaai8478 (2017). * Stuart, T. et

al. Comprehensive integration of single-cell data. _Cell_ 177, 1888–1902 (2019). * Müller, S. et al. Single-cell profiling of human gliomas reveals macrophage ontogeny as a basis for

regional differences in macrophage activation in the tumor microenvironment. _Genome Biol_. 18, 23 (2017). * Schmidt, S. V., Nino-Castro, A. C. & Schultze, J. L. Regulatory dendritic

cells: there is more than just immune activation. _Front. Immunol._ 3, 274 (2012). * Grün, D. et al. De novo prediction of stem cell identity using single-cell transcriptome data. _Cell Stem

Cell_ 19, 266–277 (2016). * Jordão, M. J. C. et al. Neuroimmunology: single-cell profiling identifies myeloid cell subsets with distinct fates during neuroinflammation. _Science_ 363,

eaat7554 (2019). * Darmanis, S. et al. Single-cell RNA-seq analysis of infiltrating neoplastic cells at the migrating front of human glioblastoma. _Cell Rep_. 21, 1399–1410 (2017). * Bunse,

L. et al. Suppression of antitumor T cell immunity by the oncometabolite (_R_)-2-hydroxyglutarate. _Nat. Med_. 24, 1192–1203 (2018). * Pusch, S. et al. Pan-mutant IDH1 inhibitor BAY 1436032

for effective treatment of IDH1 mutant astrocytoma in vivo. _Acta Neuropathol_. 133, 629–644 (2017). * Dang, L. et al. Cancer-associated _IDH1_ mutations produce 2-hydroxyglutarate. _Nature_

462, 739–744 (2009). Article  CAS  Google Scholar  * Hagos, Y. et al. Organic anion transporters OAT1 and OAT4 mediate the high affinity transport of glutarate derivatives accumulating in

patients with glutaric acidurias. _Pflugers Arch. Eur. J. Physiol_. 457, 223–231 (2008). * Mimura, J. & Fujii-Kuriyama, Y. Functional role of AhR in the expression of toxic effects by

TCDD. _Biochim. Biophys. Acta_ 1619, 263–268 (2003). * Gramatzki, D. et al. Aryl hydrocarbon receptor inhibition downregulates the TGF-β/Smad pathway in human glioblastoma cells. _Oncogene_

28, 2593–2605 (2009). Article  CAS  Google Scholar  * Silginer, M. et al. The aryl hydrocarbon receptor links integrin signaling to the TGF-β pathway. _Oncogene_ 35, 3260–3271 (2016).

Article  CAS  Google Scholar  * Gandhi, R. et al. Activation of the aryl hydrocarbon receptor induces human type 1 regulatory T cell-like and Foxp3+ regulatory T cells. _Nat. Immunol_. 11,

846–853 (2010). * Zhu, J. et al. Aryl hydrocarbon receptor promotes IL-10 expression in inflammatory macrophages through Src–STAT3 signaling pathway. _Front. Immunol._ 9, 2033 (2018).

Article  Google Scholar  * Platten, M., Nollen, E. A. A., Röhrig, U. F., Fallarino, F. & Opitz, C. A. Tryptophan metabolism as a common therapeutic target in cancer, neurodegeneration

and beyond. _Nat. Rev. Drug Discov._ 18, 379–401 (2019). * Pfefferkorn, E. R., Rebhun, S. & Eckel, M. Characterization of an indoleamine 2,3-dioxygenase induced by gamma-interferon in

cultured human fibroblasts. _J. Interferon Res_. 6, 267–279 (1986). * Pfefferkorn, E. R. Interferon γ blocks the growth of _Toxoplasma gondii_ in human fibroblasts by inducing the host cells

to degrade tryptophan (tryptophan/kynurenine/_N_-formylkynurenine). _Proc. Natl Acad. Sci. USA_ 81, 908–912 (1984). Article  CAS  Google Scholar  * Badawy, A. A. B. & Guillemin, G. The

plasma [kynurenine]/[tryptophan] ratio and indoleamine 2,3-dioxygenase: time for appraisal. _Int. J. Tryptophan Res._ 12, 1178646919868978 (2019). * Gagné, M. L., Boulay, K., Topisirovic,

I., Huot, M.-É. & Mallette, F. A. Oncogenic activities of _IDH1_/_2_ mutations: from epigenetics to cellular signaling. _Trends Cell Biol._ 27, 738–752 (2017). Article  Google Scholar  *

Friedrich, M., Bunse, L., Wick, W. & Platten, M. Perspectives of immunotherapy in isocitrate dehydrogenase-mutant gliomas. _Curr. Opin. Oncol_. 30, 368–374 (2018). * Kaper, T. et al.

Nanosensor detection of an immunoregulatory tryptophan influx/kynurenine efflux cycle. _PLoS Biol._ 5, e257 (2007). Article  Google Scholar  * Shay, T. & Kang, J. Immunological genome

project and systems immunology. _Trends Immunol._ 34, 602–609 (2013). * Rothhammer, V. & Quintana, F. J. The aryl hydrocarbon receptor: an environmental sensor integrating immune

responses in health and disease. _Nat. Rev. Immunol._ 19, 184–197 (2019). * Schmiedel, B. J. et al. Impact of genetic polymorphisms on human immune cell gene expression. _Cell_ 175,

1701–1715 (2018). * Scalise, M., Galluccio, M., Console, L., Pochini, L. & Indiveri, C. The human SLC7A5 (LAT1): the intriguing histidine/large neutral amino acid transporter and its

relevance to human health. _Front. Chem._ 6, 243 (2018). Article  Google Scholar  * Kim, C. S. et al. BCH, an inhibitor of system l amino acid transporters, induces apoptosis in cancer

cells. _Biol. Pharm. Bull_. 31, 1096–1100 (2008). * Reitman, Z. J. et al. Profiling the effects of isocitrate dehydrogenase 1 and 2 mutations on the cellular metabolome. _Proc. Natl Acad.

Sci. USA_ 108, 3270–3275 (2011). Article  CAS  Google Scholar  * Bowman, R. L., Wang, Q., Carro, A., Verhaak, R. G. W. & Squatrito, M. GlioVis data portal for visualization and analysis

of brain tumor expression datasets. _Neuro Oncol._ 19, 139–141 (2017). * Schmees, N. et al. Abstract 4454: identification of BAY-218, a potent and selective small-molecule AhR inhibitor, as

a new modality to counteract tumor immunosuppression. _Cancer Res_. https://doi.org/10.1158/1538-7445.AM2019-4454 (2019). * Kierdorf, K., Masuda, T., Jordão, M. J. C. & Prinz, M.

Macrophages at CNS interfaces: ontogeny and function in health and disease. _Nat. Rev. Neurosci._ 20, 547–562 (2019). * Friebel, E. et al. Single-cell mapping of human brain cancer reveals

tumor-specific instruction of tissue-invading leukocytes. _Cell_ 181, 1626–1642 (2020). * Klemm, F. et al. Interrogation of the microenvironmental landscape in brain tumors reveals

disease-specific alterations of immune cells. _Cell_ 181, 1643–1660 (2020). * Poon, C. C. et al. Differential microglia and macrophage profiles in human IDH-mutant and -wild type

glioblastoma. _Oncotarget_ 10, 3129–3143 (2019). Article  Google Scholar  * Sankowski, R. et al. Mapping microglia states in the human brain through the integration of high-dimensional

techniques. _Nat. Neurosci._ 22, 2098–2110 (2019). Article  CAS  Google Scholar  * Shirahata, M. et al. Novel, improved grading system(s) for IDH-mutant astrocytic gliomas. _Acta

Neuropathol_. 136, 153–166 (2018). * Geirsdottir, L. et al. Cross-species single-cell analysis reveals divergence of the primate microglia program. _Cell_ 179, 1609–1622 (2019). Article  CAS

  Google Scholar  * Tateishi, K. et al. Extreme vulnerability of _IDH1_ mutant cancers to NAD+ depletion. _Cancer Cell_ 28, 773–784 (2015). Article  CAS  Google Scholar  * Badur, M. G. et

al. Oncogenic R132 IDH1 mutations limit NADPH for de novo lipogenesis through (d)2-hydroxyglutarate production in fibrosarcoma cells. _Cell Rep_. 25, 1018–1026 (2018). * McBrayer, S. K. et

al. Transaminase inhibition by 2-hydroxyglutarate impairs glutamate biosynthesis and redox homeostasis in glioma. _Cell_ 175, 101–116 (2018). * Sinclair, L. V. et al. Control of amino-acid

transport by antigen receptors coordinates the metabolic reprogramming essential for T cell differentiation. _Nat. Immunol_. 14, 500–508 (2013). * Cibrian, D. et al. CD69 controls the uptake

of l-tryptophan through LAT1–CD98 and AhR-dependent secretion of IL-22 in psoriasis. _Nat. Immunol_. 17, 985–996 (2016). * Schumacher, T. et al. A vaccine targeting mutant IDH1 induces

antitumour immunity. _Nature_ 512, 324–327 (2014). Article  CAS  Google Scholar  * Núñez, F. J. et al. IDH1-R132H acts as a tumor suppressor in glioma via epigenetic up-regulation of the DNA

damage response. _Sci. Transl. Med_. 11, eaaq1427 (2019). * Herman, J. S., Sagar & Grün, D. FateID infers cell fate bias in multipotent progenitors from single-cell RNA-seq data. _Nat.

Methods_ 15, 379–386 (2018). * Masuda, T. et al. Spatial and temporal heterogeneity of mouse and human microglia at single-cell resolution. _Nature_ 566, 388–392 (2019). * Li, H. &

Durbin, R. Fast and accurate short read alignment with Burrows–Wheeler transform. _Bioinformatics_ 25, 1754–1756 (2009). * Baker, S. C. et al. The external RNA Controls Consortium: a

progress report. _Nat. Methods_ 2, 731–734 (2005). * Grün, D., Kester, L. & van Oudenaarden, A. Validation of noise models for single-cell transcriptomics. _Nat. Methods_ 11, 637–640

(2014). * Hafemeister, C. & Satija, R. Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression. _Genome Biol_. 20, 296 (2019).

* Bottcher, C. et al. Human microglia regional heterogeneity and phenotypes determined by multiplexed single-cell mass cytometry. _Nat. Neurosci._ 22, 78–90 (2019). Article  Google Scholar

  * Fuertig, R. et al. LC–MS/MS-based quantification of kynurenine metabolites, tryptophan, monoamines and neopterin in plasma, cerebrospinal fluid and brain. _Bioanalysis_ 8, 1903–1917

(2016). Article  CAS  Google Scholar  * Sadok, I., Rachwał, K. & Staniszewska, M. Application of the optimized and validated LC–MS method for simultaneous quantification of tryptophan

metabolites in culture medium from cancer cells. _J. Pharm. Biomed. Anal._ 176, 112805 (2019). Article  CAS  Google Scholar  * Schulte, K. W., Green, E., Wilz, A., Platten, M. & Daumke,

O. Structural basis for aryl hydrocarbon receptor-mediated gene activation. _Structure_ 25, 1025–1033 (2017). * Gibb, S. & Strimmer, K. MALDIquant: a versatile R package for the analysis

of mass spectrometry data. _Bioinformatics_ 28, 2270–2271 (2012). Article  CAS  Google Scholar  Download references ACKNOWLEDGEMENTS We are grateful to all patients who have donated tissue

for this study. We acknowledge the support of the DKFZ Light Microscopy Facility, the DKFZ Genomics and Proteomics Core Facility, the Transgenic Service of the Center for Preclinical

Research, the DKFZ and the DKFZ–Bayer Alliance. We also acknowledge support from the Flow Cytometry Core Facility at the German Cancer Research Center and the Flow Cytometry Core Facility at

the Medical Faculty Mannheim of Heidelberg University. We thank the Metabolomics Core Technology Platform at the Excellence Cluster ‘CellNetworks’ (University of Heidelberg) and the German

Research Foundation (DFG, grant ZUK 40/2010-3009262) for support with UPLC-based metabolite quantification. We acknowledge the data storage service SDS@hd supported by the Ministry of

Science, Research and the Arts Baden-Württemberg (MWK) and the German Research Foundation (DFG) through grants INST 35/1314-1 FUGG and INST 35/1503-1 FUGG. We thank D. Grün and Sagar,

MPI-IE, Freiburg, for their excellent support with processing human samples for scRNA-seq. We acknowledge the assistance of the Charité | BIH Cytometry Core (Charité, Universitätsmedizin

Berlin, Germany). We thank M. Fischer and J. Meyer for technical support. We acknowledge the support of the Freiburg Galaxy Team: R. Backofen, Bioinformatics, University of Freiburg

(Germany), funded by the Collaborative Research Centre 992 Medical Epigenetics (DFG grant SFB 992/1 2012) and the German Federal Ministry of Education and Research BMBF grant 031 A538A

de.NBI-RBC. M.F. and L.B. are members of the MD–PhD program at Heidelberg University. M.F. received fellowships from the Heidelberg Biosciences International Graduate School, the

Konrad-Adenauer Foundation, the German Academic Exchange Service (DAAD), the German Academic Scholarship Foundation (SDV) and the Excellence Initiative of the German Council of Science and

Humanities and the German Research Foundation (DFG). L.B. was funded by Heidelberg Medical Faculty and the Else Kröner Fresenius Foundation. R.S. was funded by the Berta-Ottenstein-Programme

for Clinician Scientists, Faculty of Medicine, University of Freiburg. M.K. and K.S. are supported by the Helmholtz International Graduate School. K.S. is supported by Deutsche

Forschungsgemeinschaft (DFG, German Research Foundation)—Project-ID 404521405, SFB 1389–UNITE Glioblastoma, Work Package B01. T.B. is supported by the Medical Faculty Mannheim and the

University Hospital Mannheim. J.K.S. was funded by German Cancer Aid (project 70112399). K.A. was supported by the German-Israeli Helmholtz Research School in Cancer Biology (2536). F.S. was

supported by the Else Kröner-Fresenius Excellence Program of the EKFS. C.B. was supported by the German Research Foundation (SFB TRR167 B05). J.P. was supported by the German Research

Foundation (SFB TRR167 B05 and B07, TRR265 B04) and the UK DRI Momentum Award. T.T. is supported by a fellowship from the Germany Cancer Aid non-profit organization (Deutsche Krebshilfe).

L.S. is supported by the Hertie Foundation (medMS-MyLab program; P1180016), the National Multiple Sclerosis Society (FG-1607-25111) and the Medical Faculty Mannheim, University of

Heidelberg. M. Prinz was supported by the Sobek Foundation, the Ernst Jung Foundation, the DFG (SFB 992, SFB1160, SFB1479, SFB/TRR167, a Reinhart Koselleck grant and the Gottfried Wilhelm

Leibniz prize) and the Ministry of Science, Research and Arts, Baden-Wüerttemberg (Sonderlinie ‘Neuroinflammation’). This study was supported by the DFG under Germany’s Excellence Strategy

(CIBSS, EXC-2189, project 668 ID390939984) and by the Helmholtz Gemeinschaft, Zukunftsthema ‘Immunology and Infection’ (ZT0027), the Dr. Rolf M. Schwiete Foundation and the Sonderförderlinie

‘Neuroinflammation’ of the Ministry of Science of Baden-Württemberg, the German Ministry of Education and Science (the National Center for Tumor Diseases Heidelberg NCT 3.0 program

‘Precision immunotherapy of brain tumors’ and the DKTK program), the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Project-ID 404521405, SFB 1389–UNITE Glioblastoma, Work

Package B01, and Project-ID-406052676; PL-315/5-1–Impact of dietary Tryptophan on the gut microbiome and autoimmune neuroinflammation, and the German Cancer Aid (projects 70112399 and

70113515) to M. Platten. This study was supported by grants from Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Project-ID 404521405, SFB 1389–UNITE Glioblastoma, Work

Package B01 or B03 to T.B. or L.B. and S.P., respectively, and the Medical Faculty Mannheim at Heidelberg University (“Anerkennung von Spitzenleistung” and SEED program) and the Swiss Cancer

Foundation to L.B. AUTHOR INFORMATION Author notes * These authors contributed equally: Mirco Friedrich, Roman Sankowski, Lukas Bunse, Marco Prinz, Michael Platten. AUTHORS AND AFFILIATIONS

* DKTK Clinical Cooperation Unit Neuroimmunology and Brain Tumor Immunology, German Cancer Research Center (DKFZ), Heidelberg, Germany Mirco Friedrich, Lukas Bunse, Michael Kilian, Edward

Green, Khwab Sanghvi, Markus Hahn, Theresa Bunse, Philipp Münch, Jana K. Sonner, Anna von Landenberg, Frederik Cichon, Katrin Aslan & Michael Platten * Department of Neurology,

Heidelberg University Hospital and National Center for Tumor Diseases (NCT), Heidelberg, Germany Mirco Friedrich, Lukas Bunse, Tobias Kessler & Wolfgang Wick * Department of Neurology,

MCTN, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany Mirco Friedrich, Lukas Bunse, Michael Kilian, Edward Green, Khwab Sanghvi, Markus Hahn, Theresa Bunse, Tim Trobisch, 

Lucas Schirmer & Michael Platten * Institute of Neuropathology, Faculty of Medicine, University of Freiburg, Freiburg, Germany Roman Sankowski & Marco Prinz * Faculty of Biosciences,

Heidelberg University, Heidelberg, Germany Michael Kilian, Khwab Sanghvi, Jana K. Sonner, Frederik Cichon & Katrin Aslan * Center for Mass Spectrometry and Optical Spectroscopy (CeMOS),

Mannheim University of Applied Sciences, Mannheim, Germany Carina Ramallo Guevara, Denis Abu-Sammour & Carsten Hopf * Department of Neuropathology, Heidelberg University Hospital,

Heidelberg, Germany Stefan Pusch, Daniel Schrimpf, Felix Sahm & Andreas von Deimling * DKTK Clinical Cooperation Unit Neuropathology, German Cancer Research Center (DKFZ), Heidelberg,

Germany Stefan Pusch, Daniel Schrimpf, Felix Sahm & Andreas von Deimling * Center for Organismal Studies, Heidelberg University, Heidelberg, Germany Gernot Poschet & Hagen M. Gegner

* Department of Neuroradiology, Heidelberg University Hospital, Heidelberg, Germany Philipp Münch & Sabine Heiland * DKTK Clinical Cooperation Unit Neurooncology, German Cancer Research

Center (DKFZ), Heidelberg, Germany Tobias Kessler & Wolfgang Wick * Department of Neurosurgery, University Hospital Mannheim, Mannheim, Germany Miriam Ratliff * Department of

Neurosurgery, Freiburg University Hospital, Freiburg, Germany Dieter H. Heiland, Oliver Schnell & Jürgen Beck * Department of Neuropsychiatry and Laboratory of Molecular Psychiatry,

Charité, Berlin, Germany Chotima Böttcher, Camila Fernandez-Zapata & Josef Priller * German Center for Neurodegenerative Diseases (DZNE), Berlin, Germany Josef Priller * Department of

Psychiatry and Psychotherapy, Klinikum rechts der Isar, Technical University Munich, Munich, Germany Josef Priller * University of Edinburgh and UK DRI, Edinburgh, UK Josef Priller *

Pharmaceuticals, Research and Development, Bayer AG, Berlin, Germany Ilona Gutcher * Ann Romney Center for Neurologic Diseases, Brigham and Women’s Hospital, Boston, MA, USA Francisco J.

Quintana * Signalling Research Centres BIOSS and CIBSS, University of Freiburg, Freiburg, Germany Marco Prinz * Center for Basics in NeuroModulation (NeuroModulBasics), Faculty of Medicine,

University of Freiburg, Freiburg, Germany Marco Prinz * Helmholtz Institute of Translational Oncology (HI-TRON), Mainz, Germany Michael Platten Authors * Mirco Friedrich View author

publications You can also search for this author inPubMed Google Scholar * Roman Sankowski View author publications You can also search for this author inPubMed Google Scholar * Lukas Bunse

View author publications You can also search for this author inPubMed Google Scholar * Michael Kilian View author publications You can also search for this author inPubMed Google Scholar *

Edward Green View author publications You can also search for this author inPubMed Google Scholar * Carina Ramallo Guevara View author publications You can also search for this author

inPubMed Google Scholar * Stefan Pusch View author publications You can also search for this author inPubMed Google Scholar * Gernot Poschet View author publications You can also search for

this author inPubMed Google Scholar * Khwab Sanghvi View author publications You can also search for this author inPubMed Google Scholar * Markus Hahn View author publications You can also

search for this author inPubMed Google Scholar * Theresa Bunse View author publications You can also search for this author inPubMed Google Scholar * Philipp Münch View author publications

You can also search for this author inPubMed Google Scholar * Hagen M. Gegner View author publications You can also search for this author inPubMed Google Scholar * Jana K. Sonner View

author publications You can also search for this author inPubMed Google Scholar * Anna von Landenberg View author publications You can also search for this author inPubMed Google Scholar *

Frederik Cichon View author publications You can also search for this author inPubMed Google Scholar * Katrin Aslan View author publications You can also search for this author inPubMed 

Google Scholar * Tim Trobisch View author publications You can also search for this author inPubMed Google Scholar * Lucas Schirmer View author publications You can also search for this

author inPubMed Google Scholar * Denis Abu-Sammour View author publications You can also search for this author inPubMed Google Scholar * Tobias Kessler View author publications You can also

search for this author inPubMed Google Scholar * Miriam Ratliff View author publications You can also search for this author inPubMed Google Scholar * Daniel Schrimpf View author

publications You can also search for this author inPubMed Google Scholar * Felix Sahm View author publications You can also search for this author inPubMed Google Scholar * Carsten Hopf View

author publications You can also search for this author inPubMed Google Scholar * Dieter H. Heiland View author publications You can also search for this author inPubMed Google Scholar *

Oliver Schnell View author publications You can also search for this author inPubMed Google Scholar * Jürgen Beck View author publications You can also search for this author inPubMed Google

Scholar * Chotima Böttcher View author publications You can also search for this author inPubMed Google Scholar * Camila Fernandez-Zapata View author publications You can also search for

this author inPubMed Google Scholar * Josef Priller View author publications You can also search for this author inPubMed Google Scholar * Sabine Heiland View author publications You can

also search for this author inPubMed Google Scholar * Ilona Gutcher View author publications You can also search for this author inPubMed Google Scholar * Francisco J. Quintana View author

publications You can also search for this author inPubMed Google Scholar * Andreas von Deimling View author publications You can also search for this author inPubMed Google Scholar *

Wolfgang Wick View author publications You can also search for this author inPubMed Google Scholar * Marco Prinz View author publications You can also search for this author inPubMed Google

Scholar * Michael Platten View author publications You can also search for this author inPubMed Google Scholar CONTRIBUTIONS M.F. conceptualized the study, designed and performed

experiments, analyzed and interpreted data and wrote the paper. L.B. was involved in study design and wrote the paper. R.S. analyzed scRNA-seq data and wrote the paper. M.F., M.K., J.K.S.

and K.A. performed in vivo experiments. E.G. established AHR translocation assays. G.P. performed measurements of l-Trp metabolites and TCA cycle intermediates. S.P. performed R-2-HG

measurements. T.B. was involved in study design. M.F., K.S., M.H., P.M., A.v.L. and F.C. performed in vitro experiments. T.T. and L.S. performed histological staining including RNA-ISH.

C.R.G., D.A.-S. and C.H. performed MALDI-MSI. D.H.H., O.S. and J.B. obtained informed consent from patients and dissected brain tissues. C.B., C.F.-Z. and J.P. established and performed the

CyTOF workflow for microglia as well as data processing and the clustering analysis of the CyTOF data. R.S. performed comparative analyses of CyTOF clusters. F.J.Q., W.W. and A.v.D. were

involved in study design and data interpretation. S.H. performed MRI. I.G. provided BAY-218. M. Prinz and M. Platten conceptualized the study, interpreted data and wrote the paper.

CORRESPONDING AUTHOR Correspondence to Michael Platten. ETHICS DECLARATIONS COMPETING INTERESTS M. Platten, W.W. and T.B. are inventors and patent holders on ‘Peptides for use in treating or

diagnosing IDH1-R132H positive cancers’ (EP2800580B1). S.P. and A.v.D. are eligible to royalties as co-inventors of BAY 1436032 and are patent holders of ‘Means and methods for the

determination of (D)-2-hydroxyglutarate (D2HG)’ (WO2013127997A1). The other authors declare no conflict of interest. ADDITIONAL INFORMATION PEER REVIEW INFORMATION _Nature Cancer_ thanks

Tracy McGaha and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. PUBLISHER’S NOTE Springer Nature remains neutral with regard to jurisdictional

claims in published maps and institutional affiliations. EXTENDED DATA EXTENDED DATA FIG. 1 INTEGRATED SINGLE-CELL PROFILING OF CONTROL AND HGG-DERIVED GAM. A, Donut plots showing the

cluster-wise cell compositions for the respective conditions. Clusters are arranged in a descending order from largest to smallest counterclockwise starting at 12 o’clock. Numbers indicate

clusters with significant enrichment in a given condition based on hypergeometric testing shown in Fig. 1d. N = 7 patients for control tissue, n = 5 IDHwt GBM patients, n = 5 IDHmut GBM

patients; N per condition is consistent with panel b. B, Violin plots represent probability density smoothed by a kernel density estimator with a line showing the median of corresponding

overlaid boxplots that indicate the cumulative gene expression of indicated genes as part of the antigen presentation gene expression signature in scRNA-Seq analysis. Only clusters with

significant differences in expression modelled based on a negative binomial distribution followed by pairwise testing using estimated marginal means with a significance cutoff of adjusted

p-value < 0.05 are shown. P-values were calculated using the one-way Kruskal-Wallis test followed by the Dunn post-hoc test and adjusted for multiple testing using the Benjamini-Hochberg

method. C, t-SNE map color-coded for similar clusters (obtained from the antibody panel A). D-E, t-SNE and Marimekko plot depicting the distribution of control, IDHmut and IDHwt microglia

across the clusters. F, Violin plots showing the clusterwise expression of the proteins assessed. G, Violin plots depicting the cumulative expression of ApoE and EMR1 in the respective

conditions. The color scheme is consistent with panel d. Source data EXTENDED DATA FIG. 2 INTEGRATED SINGLE-CELL PROFILING OF CONTROL AND HGG-DERIVED GAM. A, t-SNE map color-coded for

similar clusters (obtained from the antibody panel B) of n = 46,384 cells. B-C, t-SNE and Marimekko plot depicting the distribution of control (n = 16,687), IDHmut (n = 12,172) and IDHwt (n 

= 17,525) microglia across the clusters. D, Violin plots showing the clusterwise expression of the proteins assessed in the experiment by antibody panel B. E, top: t-SNE map color-coded for

the log-transformed cumulative expression of the homeostatic microglia signature proteins present in antibody panel B (CD115/CSF1R, P2RY12). The color scale indicates the color values

associated with the respective log-transformed cumulative protein expression levels. bottom: t-SNE map color-coded for the log-transformed cumulative expression of the AP signature the

protein present in antibody panel B (HLA-DR). The color scale indicates the color values associated with the respective log-transformed cumulative protein expression levels. F, Violin plots

depicting the cumulative expression of the homeostatic microglia signature proteins in the respective conditions. N = 7 control cortex tissues, n = 4 IDHmut HGG patients, n = 9 IDHwt HGG

patients. P-values were calculated using the one-way Kruskal-Wallis test followed by the Dunn post-hoc test and adjusted for multiple testing using the Benjamini-Hochberg method. The color

scheme is consistent with panel b. G, Violin plots depicting HLA-DR expression in the respective conditions. P-values were calculated using the one-way Kruskal-Wallis test followed by the

Dunn post-hoc test. Multiple testing adjustment was achieved using the Benjamini-Hochberg method. The color scheme is consistent with panels b and f. Source data EXTENDED DATA FIG. 3

ESTABLISHMENT OF AN IDH1-R132H-EXPRESSING SYNGENEIC MURINE GLIOMA MODEL. A, Vector map of scaffold/matrix attachment region (S/MAR) episomal DNA vectors encoding IDH1-R132H and IDHwt. B,

R-2-HG measurement in S/MAR-IDHwt (control), IRES-IDH1-R132H, or S/MAR-IDH1-R132H-transfected Gl261 syngeneic murine glioma cell lines. Statistical significance was determined by two-tailed

student’s t test. Cell lines were sampled and examined over 4 independent measurements. Data are represented as mean ± SEM. C, Immunofluorescence staining of IDH1-R132H (green) in

S/MAR-IDHwt (control) or S/MAR-IDH1-R132H-transfected Gl261 syngeneic murine glioma cell line. Staining was performed once for each cell line shown in b. D, R-2-HG measurement in explanted

intracranial tumors of GL261-S/MAR-IDH1-R132H-bearing mice. Median concentration indicated as broken line. N = 5 tumor-bearing mice. E, Inhibition of mutant IDH is efficient in a syngeneic

murine glioma model by reverting R-2-HG-induced TAM phenotype. N = 7 IDHwt tumor-bearing mice, n = 7 IDHmut tumor-bearing mice + vehicle, n = 4 IDHmut tumor-bearing mice + BAY 1436032. Tumor

volumes measured by T2-weighted MRI. Statistical significance was determined by one-way ANOVA in combination with Tukey’s test. F, Bulk RNA-Sequencing of GL261-S/MAR-IDH1-wildtype (IDHwt)

and GL261-S/MAR-IDH1-R132H (IDHmut) cell lines indicating genes that have been previously highlighted to indicate mesenchymal and proneural glioma phenotypes, respectively42,53. The color

scale indicates z-scores. G, MALDI-MS imaging of R-2-HG in snap-frozen tissue of explanted intracranial GL261 tumors expressing S/MAR-IDH1-R132H or S/MAR-IDHwt. Left: Exemplary heatmap of

measured R-2-HG intensity, right: normalized R-2-HG in arbitrary units (AU) shown. Data points aggregated from n = 6 independent explanted tumors. Box and whiskers (10–90 percentile, median

as center) shown. H-I, Pseudotime analysis of the stepwise changes between clusters enriched for cells of early stage tumors (H) and late stage tumors (I) respectively. Data is suggested by

the StemID2 algorithm generating a GAM lineage tree. N = 14 samples from n = 4 IDHwt tumor-bearing mice (d7), n = 4 IDHmut tumor-bearing mice (d7), n = 3 IDHwt tumor-bearing mice (d28), n = 

3 IDHmut tumor-bearing mice (d28). Source data EXTENDED DATA FIG. 4 R-2-HG-DRIVEN AHR SIGNALING OF GAM. A, Intracellular measurements of R-2-HG in primary microglia after _in vitro_

incubation with R-2-HG for 24 h. Nonlinear regression is shown. N = 1 healthy donor. B, Intracellular measurements of R-2-HG in human MΦ. Cells were primed for 2 h with 100 μM Candesartan,

100μM N-(p-Amylcinnamoyl)anthranilic acid (NAA) or vehicle only and treated with 20mM R-2-HG for 48 h. N = 4 healthy donors. Data are represented as mean ± SEM. Statistical significance was

determined by one-way ANOVA in combination with Tukey’s test. C, DNA-microarray screen of MΦ from n = 8 healthy donors, treated with exogenous R-2-HG in a matched-pair analysis: procedural

overview. D, Quantification of AHR signature expression in IDHwt (grey) and IDHmut (blue) LGG (absolute (left) and normalized to CD11b- (ITGAM-) expression (right)). TCGA dataset, N = 286, n

 = 68 IDHwt LGG, n = 218 IDHmut LGG. E, AHR target gene expression in human monocyte-derived MΦ treated with increasing concentrations of R-2-HG _in vitro_ as determined by RT-PCR. N = 4

healthy donors. Statistical significance was determined by one-way ANOVA in combination with Tukey’s test. F, Expression levels of AHR in transcripts per million (TPM) in human immune cell

subsets as analyzed in the Database of Immune Cell Expression (DICE). G, Cytokine ELISA measurements of IL-10 and TGF-β in human monocyte-derived MΦ following incubation with increasing

doses of R-2-HG. Statistical significance was determined by two-tailed student’s t test. N = 4 independent healthy donors evaluated over 6 experimental conditions H. Induction of AHR

translocation by supernatant of IDH1-R132H-expressing glioma cells (blue). Supernatant of IDH1-wt-overexpressing GL261 glioma cell line as control (grey). N = 3 IDHwt cell lines, n = 3

IDHmut cell lines. Statistical significance was determined by two-tailed student’s t test. I, AHR translocation reporter assay. Titration of R-2-HG and L-Kyn in DRE-GFP-reporter, n = 3 assay

runs. Statistical significance was determined by one-way ANOVA in combination with Tukey’s test. Source data EXTENDED DATA FIG. 5 LAT1-CD98-DEPENDENT L-TRP UPTAKE OF GAM. A, Intracellular

L-Trp measurement of R-2-HG-treated MΦ from n = 5 healthy human donors by UPLC-FLR/UV. Fold change R-2-HG/vehicle given on the right. Statistical significance was determined by two-tailed

student’s t test. B-C, Multiplex fluorescence _in situ_ hybridization of IDHwt, IDHmut glioma as well as control cortical tissue, showing the expression of CD163 (green), SLC7A5 (red), CD44

(white) and nuclear marker DAPI (blue). White arrowheads mark SLC7A5-expressing macrophages, grey arrowheads mark reactive astrocytes expressing SLC7A5. Staining was repeated with 5

independent patient samples and 1 control cortex tissue. Source data EXTENDED DATA FIG. 6 ESTABLISHING A MALDI-MSI APPROACH FOR DETECTION OF R-2-HG-INDUCED MICROENVIRONMENTAL CHANGES IN

PRIMARY HUMAN TISSUE. A, 1, Cryosectioning and mounting of fresh frozen tissue specimens. Cutting scheme: (A) the first 10 µm tissue section of each specimen was mounted onto a conductive

ITO slide for MALDI-MSI while (B) the adjacent tissue sections (10 µm) were then mounted on a Superfrost© microscope glass slide for standard histological analysis. 2, MALDI matrix

deposition using an automated Sprayer. Frozen tissue sections on ITO slides were spray-coated with 1,5-diaminonaphtalene (1,5-DAN) prepared at 10 mg/mL with 50% ACN (v/v) matrix for analyte

extraction and ionization. 3, Pixel-wise MALDI MS analysis using a high-resolution mass spectrometer. Molecular ions generated after laser ablation are separated by their mass to charge

ratio. At each pixel a mass spectrum is recorded in a grid like manner for multiple (x,y)-coordinates for data acquisition. 4, Data processing was performed with an in-house developed

pre-processing pipeline using the R programming environment. 5, Image reconstruction & histological verification. For each selected m/z (z = 1) value the intensity map is shown as a

color-coded distribution map. Merely one ion image per mass can be reconstructed. H&E stained image confirms the tumor sites within the tissue section. 6, Data analysis to estimate the

levels of detected intensities of the analytes of interest (m/z) in different tissue specimens. B, Local polynomial regression analysis plot for L-Trp as a function of increasing signal

intensity of R-2-HG when both are co-localized. A locally estimated scatterplot smoothing (LOESS) curve was fitted which indicates that L-Trp exhibits a general tendency of increasing signal

intensity when R-2-HG is increasing. SEM is projected as error bands. N = 5 control cortex samples, n = 5 IDHmut patient samples, n = 5 IDHwt patient samples C. AHR reporter assay after

treatment with increasing doses of R-2-HG + vehicle, R-2-HG after AHRi pretreatment (1 h) or PBS for 6 h. N = 3 independent experiments. SEM is projected as error bands. If not mentioned

otherwise, all data are represented as mean ± SEM. Source data SUPPLEMENTARY INFORMATION SUPPLEMENTARY INFORMATION Supplementary Figs. 1 and 2. REPORTING SUMMARY SUPPLEMENTARY TABLES

Supplementary Tables 1–8 SOURCE DATA SOURCE DATA FIG. 1 Statistical source data. SOURCE DATA FIG. 2 Statistical source data. SOURCE DATA FIG. 3 Statistical source data. SOURCE DATA FIG. 4

Statistical source data. SOURCE DATA FIG. 5 Statistical source data. SOURCE DATA FIG. 6 Statistical source data. SOURCE DATA FIG. 7 Statistical source data. SOURCE DATA EXTENDED DATA FIG. 1

Statistical source data. SOURCE DATA EXTENDED DATA FIG. 2 Statistical source data. SOURCE DATA EXTENDED DATA FIG. 3 Statistical source data. SOURCE DATA EXTENDED DATA FIG. 4 Statistical

source data. SOURCE DATA EXTENDED DATA FIG. 5 Statistical source data. SOURCE DATA EXTENDED DATA FIG. 6 Statistical source data. RIGHTS AND PERMISSIONS OPEN ACCESS This article is licensed

under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate

credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article

are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and

your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this

license, visit http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Friedrich, M., Sankowski, R., Bunse, L. _et al._ Tryptophan

metabolism drives dynamic immunosuppressive myeloid states in IDH-mutant gliomas. _Nat Cancer_ 2, 723–740 (2021). https://doi.org/10.1038/s43018-021-00201-z Download citation * Received: 15

July 2020 * Accepted: 18 March 2021 * Published: 24 May 2021 * Issue Date: July 2021 * DOI: https://doi.org/10.1038/s43018-021-00201-z SHARE THIS ARTICLE Anyone you share the following link

with will be able to read this content: Get shareable link Sorry, a shareable link is not currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt

content-sharing initiative