Visual Abstract

Fumarate is a tricarboxylic acid cycle metabolite whose intracellular accumulation is linked to inflammatory signaling and development of cancer. In this study, we demonstrate that endogenous fumarate accumulation upregulates surface expression of the immune stimulatory NK group 2, member D (NKG2D) ligands ULBP2 and ULBP5. In agreement with this, accumulation of fumarate by the therapeutic drug dimethyl fumarate (DMF) also promotes ULBP2/5 surface expression. Mechanistically, we found that the increased ULBP2/5 expression was dependent on oxidative stress and the antioxidants N-acetylcysteine and glutathione (GSH) abrogated ULBP2/5 upregulated by DMF. Fumarate can complex with GSH and thereby exhaust cells of functional GSH capacity. In line with this, inhibition of GSH reductase (GR), the enzyme responsible for GSH recycling, promoted ULBP2/5 surface expression. Loss of the tricarboxylic acid cycle enzyme fumarate hydratase (FH) associates with a malignant form of renal cancer characterized by fumarate accumulation and increased production of reactive oxygen species, highlighting fumarate as an oncometabolite. Interestingly, FH-deficient renal cancer cells had low surface expression of ULBP2/5 and were unresponsive to DMF treatment, suggesting that the fumarate-stimulating ULBP2/5 pathway is abrogated in these cells as an immune-evasive strategy. Together, our data show that ULBP2/5 expression can be upregulated by accumulation of fumarate, likely by depleting cells of GSH antioxidant capacity. Given that DMF is an approved human therapeutic drug, our findings support a broader use of DMF in treatment of cancers and inflammatory conditions.

An emerging hallmark of inflammation and cancer is the modification of metabolic enzymes that results in a changed metabolic phenotype (1). The citric acid cycle oxidizes lipids, amino acids, and carbohydrates into CO2 and ATP. The citric acid cycle can, however, also function as a hub for production of anabolic precursors. In this way, activated macrophages, T cells, and different types of cancer cells associate with a breakage of the citric acid cycle that enables the use of citric acid cycle intermediates for stimulatory functions and proliferation (25).

Fumarate is a citric acid cycle metabolite that is generated from succinate by the enzyme succinate dehydrogenase (SDH). Accumulation of fumarate has previously been observed in LPS-activated monocytes, and recently, fumarate has emerged as an inflammatory signal that regulates trained immunity through epigenetic reprogramming of β-glucan–treated monocytes (6, 7). In addition to its role in inflammation, accumulated fumarate is linked to transformation (8, 9). Turnover of fumarate into malate is mediated by fumarate hydratase (FH), an enzyme characterized as a tumor suppressor gene whose inactivation is responsible for development of hereditary leiomyomatosis renal cell carcinoma (HLRCC) (10). Loss of FH activity in HLRCC is associated with citric acid cycle breakage and chronic accumulation of fumarate that amplifies reactive oxygen species (ROS)–dependent signaling, emphasizing fumarate as an oncometabolite (9, 11, 12).

Inflammation and transformation can be detected by the immune system, which responds to abnormal cells and maintains homeostasis. The NK group 2, member D (NKG2D) is an immune-activating receptor primarily expressed by NK cells and different subsets of T cells (13, 14). It recognizes NKG2D ligands that are self-proteins upregulated on the surface of stressed, transformed, or infected cells. NKG2D ligands belong to the MHC class I–related family (MHC class I polypeptide-related sequence [MIC] A/B) or the UL16-binding protein family (ULBP1-6) (15, 16). Different cellular stresses, including heat shock, histone deacetylase (HDAC) inhibitors, and DNA damage responses, result in NKG2D ligand surface expression (1719). Furthermore, NKG2D ligands are induced by propionate, a short chain fatty acid (SCFA) and high glucose, indicating that metabolic stress also regulates NKG2D-mediated immunity (20, 21). In addition to transcriptional regulation, surface expression of NKG2D ligands can be regulated by posttranslational mechanisms [e.g., pathogen-mediated change of MICA N-glycosylation and invariant chain–dependent endosomal transport of ULBP2 (2224)].

The NKG2D system is central for sensing and responding to several stress conditions, making it an attractive target to avoid immune recognition. Several viruses and cancers have been linked to obstruction of functional NKG2D ligand surface expression through intracellular retention and cleavage from the cell surface (2529), most likely serving as immune-evasive strategies.

The antioxidant glutathione (GSH) system counteracts cellular oxidative stress and is involved in redox homeostasis, detoxification, and antioxidant defense (30). Mammalian cells synthesize GSH in the cytosol from the amino acids glutamine, cysteine, and glycine (31). GSH exists in two forms, the thiol-reduced GSH and the disulfide-oxidized l-GSH (GSSG) (32). In the antioxidant defense, GSH is oxidized to the inactivated disulfide form GSSG. De novo synthesis of GSH is limited by the availability of cysteine and the activity of the rate-limiting enzyme glutamate cysteine ligase (GCL) (33). Recycling of GSSG to reduced GSH is essential to maintain functional GSH concentrations within cells. The enzyme GSH reductase (GR) facilitates this process while consuming NADPH (30). GR activity is therefore crucial to maintain adequate GSH antioxidant capacity. Mitochondrial citric acid cycle activity and oxidative phosphorylation produce ROS involved in normal homeostatic cell signaling (34, 35). However, excessive ROS production is harmful and causes oxidative stress that promotes dysregulated proliferation (1, 36). In this respect, the GSH antioxidant system is vital for neutralizing ROS activity during oncogenic proliferation.

Dimethyl fumarate (DMF) is approved as a therapeutic drug for treatment of the autoimmune diseases multiple sclerosis and psoriasis (37, 38). Despite widespread usage of DMF in the clinic, the mode of action is currently elusive. Proposed mechanisms include activation of the antioxidant transcription factor Nrf2 as well as inhibition of the inflammatory NF NF-κB (37, 3941). Furthermore, DMF and fumarate covalently modify cysteine residues. Both in vitro and in vivo studies have shown that fumarate binds directly to thiol groups present in GSH, thereby producing the metabolite succinated GSH (GSF) that modulates redox sensitivity and inflammatory pathways (9, 42, 43). DMF has also been shown to inhibit growth and metastasis of melanoma cells (44).

In this study, we found that inhibition of FH activity upregulated surface expression of the NKG2D ligands ULBP2 and ULBP5. Our data indicate that the upregulation was caused by accumulation of fumarate because treatment with DMF also upregulated ULBP2/5 surface expression. We further show that the antioxidants N-acetylcysteine (NAc) and GSH completely abrogated expression of ULBP2/5 regulated by DMF. In line with this, inhibition of GR activity promoted a robust ULBP2/5 surface expression in the absence of DMF, indicating the importance of diminished GSH antioxidant capacity for ULBP2/5 expression. Interestingly, FH-deficient renal cancer cells with elevated levels of fumarate had low basal ULBP2/5 surface expression that was not regulated by DMF treatment, suggesting that FH-deficient cells with pathological accumulation of fumarate have selectively hampered the fumarate–ULBP2/5 stimulatory pathway as an immune-evasive strategy.

Together, our data highlight a novel mechanism in which accumulation of fumarate depletes cells from GSH, ultimately leading to ULBP2/5 surface expression. Many cancer cells show an elevated level of GSH that enables them to cope with high ROS production and resist chemotherapy (31, 45). The GSH antioxidant system, therefore, has the potential to be a major target for treatment of cancer. Our data highlight DMF as a novel anticancer drug that can target transformed cells to NKG2D-mediated killing by exhausting GSH antioxidant capacity.

Jurkat TAg9 T cells (Jurkat cells) were kindly provided by Professor Carsten Geisler (Department of Immunology and Microbiology, University of Copenhagen, Copenhagen, Denmark). Jurkat cells are stably transfected with the large TAg from SV40. The prostate adenocarcinoma cell line PC3 and the human acute monocytic leukemia cell line THP1 were purchased from American Type Culture Collection. 2B4 and 2B4_NKG2D cell lines were kindly provided by Dr. Chiwen Chang (Department of Pathology, University of Cambridge, Cambridge, U.K.). The FH-deficient cell line UOK262 was kindly provided by Dr. W. Marston Linehan (Center for Cancer Research, National Cancer Institute, Bethesda, MD). The FH-deficient cell line NCCFH1 was kindly provided by Dr. Min Han-Tan (Institute of Bioengineering and Nanotechnology, The Nanos, Singapore) and Dr. Bin Tean Teh (Laboratory of Cancer Epigenome, National Cancer Center, Singapore). Jurkat, THP1 and PC3 cell lines were cultivated in RPMI 1640 medium (R5886; Sigma-Aldrich) supplemented with 2 mM l-glutamine (G7513; Sigma-Aldrich), 2 mM penicillin/streptomycin (P4333; Sigma-Aldrich) and 10% v/v FBS (F9665; Sigma-Aldrich). The HEK293, UOK262, and NCCFH1 cell lines were cultivated in DMEM with GlutaMAX (31966047; Thermo Fisher Scientific) supplemented with 2 mM penicillin/streptomycin and 10% v/v FBS. Buffy coats from healthy human donors were obtained from Capital Region Blood Bank, Copenhagen University Hospital, Copenhagen. PBMCs were isolated by density gradient using Histopaque-1077 (10771; Sigma-Aldrich). Monocytes were negatively isolated from PBMCs using Pan Mouse IgG Dynabeads (11042; Invitrogen), as previously described (46). The remaining PBLs were cultivated in RPMI medium supplemented with 20 U/ml IL-2 (200-02; PeproTech) and activated with CD3/CD28 Dynabeads (11132D; Thermo Fisher Scientific) for 3 d before treatment with DMF. NK cells were positively isolated from PBMCs using NKp46 Ab (16-3359-82; Invitrogen) and a Dynabeads FlowComp Flexi Kit (11061D; Invitrogen) according to the manufacturer’s protocol. NK cells were treated immediately with DMF or activated for 3 d with 500 U/ml IL-2 or 10 ng/ml IL-15 (200-15; PeproTech) before DMF treatment.

The HDAC inhibitor FR901228 was kindly provided by the National Cancer Institute (Bethesda, MD). DMF (242926), NAc (A-9165), GSSG (G-4501), reduced l-GSH (GSH) (G-4251), 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) (C0400), sodium propionate (P1880), PBS (D8537), 6-aminonicotinamide (6-AN) (A68203), and DMSO (D2438) were all purchased from Sigma-Aldrich. Cell-permeable succinate prodrug (CP-Succ) (NV-01-118) and SDH inhibitor (SDHi; NV-01-161) were purchased from Isomerse Therapeutics. FH-IN-1 (FHi) (1644060-37-6) was purchased from MedChem.

Jurkat cells and PBLs were plated to a density of 6 × 105 cells/ml before stimulation with the indicated concentrations of FHi, DMF, or CP-Succ at 37°C, 5% CO2 for 18–48 h. The HDAC inhibitor FR901228 and the SCFA propionate were used as positive controls for NKG2D ligand induction. For vehicle controls, an equal volume of DMSO or PBS was used. For the antioxidant experiments, Jurkat cells were pretreated with 100 μM GSH, 100 μM GSSG, or 5 mM NAc before being treated with DMF, CP-Succ, or FR901228. For the GR inhibition, Jurkat cells were treated with the indicated concentrations of BCNU either alone or in combination with 100 μM GSH or 5 mM NAc. PC3 cells were plated to a density of 5 × 104 cells/ml and allowed to attach overnight (o/n) in the incubator before being treated with FHi, DMF, or CP-Succ. NCCFH-1 and UOK262 cells were plated to a density of 4–6 × 104cells/ml and allowed to attach o/n in the incubator before being treated with DMF, CP-Succ, FR901228, propionate, BCNU, or 6-AN. To detect intracellular ROS, Jurkat cells were treated with 30 μM DMF, 70 μM CP-Succ, 80 μM BCNU, or 100μM 6-AN for the indicated time before being stained at 37°C for 30 min with 1.25 μM MitoSOX (M36008; Thermo Fisher Scientific) or 5 μM 2′7′-dichlorodihydrofluorescein diacetate (H2DCFDA) (D399; Thermo Fisher Scientific) dissolved in RPMI medium. The cells were washed twice in PBS supplemented with 2% v/v FBS and analyzed by flow cytometry.

For the reporter cell assay, target cells (THP1) were plated to a density of 3 × 105 cells/ml and treated with DMSO or DMF for 24 h. Surface expression of ULBP2/5/6 was verified by flow cytometry. Prior to cocultivation, target cells were washed, whereas 2B4 and 2B4_NKG2D (effector cells) were labeled with the Vybrant cell dye DiD (V22889; Invitrogen) according to the manufacturer’s protocol. Target and effector cells were mixed at the indicated ratio and cocultivated overnight at 37°C, 5% CO2. GFP expression (NKG2D receptor activation) was measured by flow cytometry. For NKG2D blocking, THP1 cells were washed and incubated with NKG2D–Fc (1299-NK; R&D Systems) or control IgG1–Fc (110-HG; R&D Systems) for 30 min at 4°C before being cocultivated with the effector cells.

For the NKG2D down-modulation assay, target THP1 cells were treated as in the reporter cell assay. CD4+ T cells were positively depleted from PBMCs using CD4 Ab (16–0049; eBioscience) and Dynabeads (11041; Invitrogen). CD4+-depleted PBLs (effector cells) were activated for 3 d with 10 ng/ml IL-15. NK cells were purified from buffy coats using RosetteSep Human NK Cell Enrichment Cocktail (15065; STEMCELL Technologies) according to the manufacturer’s protocol. NK cells (effector cells) were activated o/n with 500 U/ml IL-2 and 10 ng/ml IL-15. Before cocultivation, target THP1 cells were washed and incubated with NGK2D–Fc or control IgG1–Fc for 30 min at 4°C. Target and effector cells were mixed at the indicated ratio and cocultivated for 2–3.5 h at 37°C, 5% CO2 before NKG2D expression was analyzed on CD4+-depleted PBLs or CD56+/CD3 NK cells. NKG2D down-modulation on CD4+-depleted PBLs or CD56+/CD3 NK cells was assessed by NKG2D surface expression blocked with control IgG1–Fc relative to NKG2D–Fc.

Suspension cells were resuspended, and 200 μl of cell culture was transferred to a 96-well FACS plate. Adherent cells were detached by incubation at 37°C in versene (15040-033; Life Technologies) for 30 min before they were resuspended and transferred to a 96-well FACS plate. After centrifugation at 300 × g for 3 min at 4°C, cells were washed twice in PBS supplemented with 2% v/v FBS and stained with the indicated Ab for 15 min on ice. THP1 monocytes were blocked with Fc receptor–blocking solution (120-000-442; Miltenyi Biotec) for 10 min on ice prior to the specific Ab stain. After Ab staining, cells were washed twice in PBS with FBS and transferred to FACS analysis tubes. The following Abs were used: PE–IgG2a (555574), allophycocyanin–IgG2a (555576,), allophycocyanin­–IgG1 (555751), PE–IgG2b (555743), FITC–IgG2b (555742), PE–MICA/B (558352), PE–CD3 (555340), FITC–CD56 (543811) (all purchased from BD Biosciences); allophycocyanin–ULBP2/5/6 (FAB1298A), PE–ULBP1 (FAB1380P), allophycocyanin–ULBP3 (FAB1517A), PE–ULBP4 (FAB6285P), allophycocyanin–NKG2D (FAB139A) (all purchased from R&D Systems), allophycocyanin–annexin V (640919), and allophycocyanin–anti-mouse IgG1 (405308) (purchased from BioLegend); and anti-myc tag (05-724; Merck Millipore). Propidium iodide (P4864) was purchased from Sigma-Aldrich. The soluble NKG2D–Fc receptor (1299-NK, R&D Systems) and IgG1–Fc (110-HG; R&D Systems) were labeled with Zenon Alexa Fluor 647 against human IgG1 (Z25408; Invitrogen) prior to staining of Jurkat cells. Flow cytometry data were acquired on a BD Accuri C6 flow cytometer using the BD Accuri C6 Software. Data analysis was performed using FlowLogic 700.2A Software (Inivai Technologies). All samples were analyzed by gating on forward and side scatter. The grid is set relative to 5% positive cells in the vehicle-treated sample if otherwise not specified. Bar graphs show mean fluorescence intensities subtracted from the respective isotype control or the percentage of positive cells relative to 5% positive in the vehicle-treated sample if otherwise not specified.

Jurkat cells were treated with DMSO, DMF, CP-Succ, FR901228, or propionate for 18 h before being lysed 30 min on ice in RIPA lysis buffer (89900; Thermo Fisher Scientific) supplemented with 100× Protease/Phosphatase Inhibitor Cocktail (1861281; Thermo Fisher Scientific). Cell lysates were mixed with 4× NuPAGE sample buffer (NP0007; Invitrogen) and 20× DTT (646563, Sigma-Aldrich) before being incubated at 70°C for 10 min. Proteins were separated in a NuPAGE Novex 4–12% Bis-Tris Midi Protein Gel (WG1401; Invitrogen) with 1× NuPAGE MOPS SDS Running Buffer (NP0001; Invitrogen) and antioxidant (NP005; Invitrogen) for 1 h 15 min at 200V. An iBlotTM Gel Transfer Device was used to transfer proteins to a nitrocellulose membrane (IB301001; Invitrogen). The membrane was blocked in PBS blocking buffer (927–40000, LI-COR Biosciences) for 45 min before being stained with anti-p21 (2947S; Cell Signaling Technology) and anti-GAPDH (A01622; GenScript) at 4°C overnight. The membrane was washed in PBS plus 0.1% Tween 20 (P9416; Sigma-Aldrich) and stained with IRDye secondary Abs: anti-rabbit IgG (926–32213, LI-COR Biosciences) and anti-mouse IgG (926–68022, LI-COR Biosciences) for 1 h. After washing in PBS plus 0.1% Tween 20, the membrane was developed using the LI-COR Odyssey Fc instrument.

Cells were treated with the indicated compounds for 4 h before total RNA was isolated using TRIzol reagent (15596026; Invitrogen), chloroform (C2432; Sigma-Aldrich), and Direct-zol RNA MiniPrep (R2050S; Zymo Research). Quanta qScript cDNA SuperMix (95048; Quanta Biosciences) was used to reverse transcribe 1 μg of RNA into cDNA. Primer sequences for ULBP2 were as follows: Forward_ULBP2, 5′-CAGAGCAACTGCGTGACATT-3′ and Reverse_ULBP2, 5′-GGCCACAACCTTGTCATTCT-3′. For MICA, the sequences were as follows: Forward_MICA, 5′-GCCATGAACGTCAGGAATTT-3′ and Reverse_MICA, 5′-GACGCCAG-CTCAGTGTGATA-3′. For ULBP5, the sequences were as follows: Forward_ULBP5, 5′-CTGCGATCCAACTCCCCAATG-3 ′ and Reverse_ULBP5, 5′-GGTCCAGGTCTGAACTTAGGG-3′. For ULBP6, the sequences were as follows: Forward_ULBP6, 5′-TTCAT-CTTCCAGGATCCACCTT-3′ and Reverse_ULBP6, 5′-CGTGGTCCAGGTCTGAACTT-3′. The ribosomal protein, large P0 (RPLP0) was included as a housekeeping gene. Primer sequences for RPLP0 were as follows: Forward_RPLP0, 5′-CCTCGTGGAAGTGACATCGT-3′ and Reverse_ RPLP0, 5′-CA-TTCCCCCG-GATATGAGGC-3′. All primers were purchased from Eurofins MWG Operon. Quantitative real-time PCR was performed using SYBR Green Master Mix (208056; QuantiNova) on an Agilent AriaMx apparatus. Analysis was done using Agilent AriaMx Software (Agilent, Santa Clara, CA). Relative expression of ULBP2/5 and MICA was calculated using the 2−ΔΔCt method (versus DMSO).

Transient transfection of FH-deficient UOK262 cells was performed with the Amaxa Nucleofector device (Lonza). Briefly, 3 × 105 UOK262 cells were resuspended in 100 μl Nucleofector solution V (VCA-1003; Lonza) and mixed with 1 μg of pmaxGFP (Lonza) or 1 μg of myc-tagged GFP–ULBP2 [previously described (22)] before being pulsed using the Nucleofector program A-024. Expression of plasmids was measured by flow cytometry after 24 h.

Graphs and analysis were done using GraphPad Prism version 8.0 (GraphPad Software). Statistical analysis was performed using an unpaired Welch t test (Figs. 1A–C, 1E–H, 2A–C, 3A–C, 4B, 4F, 5A, 5B, 6B, 6C, 6E, 6F, Supplemental Figs. 1A, 1B, 1F–I, 3D–F) or two-way ANOVA with Bonferroni multiple comparison test (Fig. 1I, 1J, 3D–F, 4D, 4E, 5C, 5E, 7A, 7G, 7L, 7M, Supplemental Figs. 3B, 3C). Data are presented as mean ± SD. For all experiments, the level of statistical significance was set at p < 0.05.

FH converts fumarate to malate in the citric acid cycle under normal physiological conditions. Regulation of FH activity is therefore vital for fumarate turnover and mitochondrial respiration. This is highlighted in FH-deficient renal cell carcinomas, in which loss of FH activity associates with fumarate accumulation and increased ROS production (9, 11). Other cellular changes related to fumarate accumulation include dysregulated metabolism, epigenetic changes in immune cells, and activation of stress pathways (7, 8).

To explore the immune regulatory functions of fumarate, we examined whether accumulation of fumarate-regulated surface expression of the stress-induced NKG2D ligands ULBP2/5/6 and MICA/B. We treated Jurkat cells with the FHi, which blocks the enzymatic activity of FH and thereby causes accumulation of fumarate. Inhibition of FH activity resulted in surface expression of ULBP2/5/6 on Jurkat cells after 48 h as measured by flow cytometry (Fig. 1A). We did not observe an effect of FH inhibition on MICA/B surface expression (Supplemental Fig. 1A). The data indicate that endogenous accumulation of fumarate generates metabolic stress that is sufficient to upregulate ULBP2/5/6 but not MICA/B surface expression.

FIGURE 1.

Accumulation of fumarate upregulates surface expression of ULBP2/5/6. (A) Jurkat cells were treated with vehicle control (DMSO) or the indicated concentrations of FHi for 48 h before being analyzed for surface expression of ULBP2/5/6 by flow cytometry. (BD) Jurkat cells were treated with the indicated concentrations of DMF for 18 h and analyzed for ULBP2/5/6 and MICA/B surface expression. FR901228 (20 ng/ml) was included as a positive control. (B) Representative dot plots of Jurkat cells treated with DMSO, DMF, or FR901228. (E) Human monocytic THP1 cells were treated with DMSO or 100 μM DMF for 24 h before being analyzed for surface expression of ULBP2/5/6. (F) Prostate cancer PC3 cells were treated with DMSO or 80 μM DMF for 18 h before being analyzed for surface expression of ULBP2/5/6. (G) PBLs purified from buffy coats from three healthy donors were treated with DMSO or 40 μM DMF for 18 h and analyzed for surface expression of ULBP2/5/6. (H) NK cells from healthy donors were treated with the indicated concentrations of DMF at time of purification (resting) or stimulated for 3 d with IL-2 (500 U/ml) and IL-15 (10 ng/ml) before DMF treatment. ULBP2/5/6 and CD56 surface expression were analyzed by flow cytometry. (I) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 either alone or in combination with 5 μM FHi. (J) PC3 cells were treated with DMSO or 80 μM DMF either alone or in combination with 5 μM FHi. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 1.

Accumulation of fumarate upregulates surface expression of ULBP2/5/6. (A) Jurkat cells were treated with vehicle control (DMSO) or the indicated concentrations of FHi for 48 h before being analyzed for surface expression of ULBP2/5/6 by flow cytometry. (BD) Jurkat cells were treated with the indicated concentrations of DMF for 18 h and analyzed for ULBP2/5/6 and MICA/B surface expression. FR901228 (20 ng/ml) was included as a positive control. (B) Representative dot plots of Jurkat cells treated with DMSO, DMF, or FR901228. (E) Human monocytic THP1 cells were treated with DMSO or 100 μM DMF for 24 h before being analyzed for surface expression of ULBP2/5/6. (F) Prostate cancer PC3 cells were treated with DMSO or 80 μM DMF for 18 h before being analyzed for surface expression of ULBP2/5/6. (G) PBLs purified from buffy coats from three healthy donors were treated with DMSO or 40 μM DMF for 18 h and analyzed for surface expression of ULBP2/5/6. (H) NK cells from healthy donors were treated with the indicated concentrations of DMF at time of purification (resting) or stimulated for 3 d with IL-2 (500 U/ml) and IL-15 (10 ng/ml) before DMF treatment. ULBP2/5/6 and CD56 surface expression were analyzed by flow cytometry. (I) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 either alone or in combination with 5 μM FHi. (J) PC3 cells were treated with DMSO or 80 μM DMF either alone or in combination with 5 μM FHi. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Close modal

We next investigated whether fumarate accumulation through DMF treatment upregulated surface expression of ULBP2/5/6 as well as other ULBP proteins. Similar to FH inhibition, DMF upregulated surface expression of ULBP2/5/6 in a concentration-dependent manner to a level comparable with the HDAC inhibitor FR901228, a well-known inducer of NKG2D ligands (Fig. 1B, 1D) (17). Moreover, DMF upregulated surface expression of ULBP1 but not ULBP3 and ULBP4 (Supplemental Fig. 1B). Because Jurkat cells express a high basal level of ULBP2/5/6, the grid was set relative to DMSO-treated cells and not relative to the isotype control (Supplemental Fig. 1C). In contrast to FH inhibition, DMF also upregulated surface expression of MICA/B but not as potently as the HDAC inhibitor FR901228 (Fig. 1C, 1D). To minimize cell death, we used 30 μM DMF as the standard DMF concentration in Jurkat cells (Supplemental Fig. 1D). In line with this, increased ULBP2/5/6 surface expression was only observed on annexin V–nonapoptotic cells (Supplemental Fig. 1E). In the rest of the study, we focused on ULBP2/5/6 and MICA/B as examples of different NKG2D ligands. We assessed the kinetics of ULBP2/5/6 and MICA/B upregulated by DMF and found surface expression of these NKG2D ligands after 12 and 18 h, respectively (Supplemental Fig. 1F, 1G).

We examined different cell types in response to DMF and found increased surface expression of ULBP2/5/6 on the leukemic monocyte THP1 cell line (Fig. 1E), the prostate cancer cell line PC3 (Fig. 1F), and CD3/CD28–activated human PBLs (Fig. 1G). Notably, DMF did not increase ULBP2/5/6 on resting NK cells or NK cells activated with IL-2/IL-15 (Fig. 1H). Cancer cells often engage Warburg metabolism, characterized by aerobic glycolysis and lactate production, a metabolism that is shared by CD3/CD28–activated CD4+ and CD8+ T cells (1, 47). In contrast to the tested cancer cell lines and activated PBLs, DMF did not upregulate ULBP2/5/6 and MICA/B on noncancer HEK293 cells (Supplemental Fig. 1H, 1I).

To investigate the combined effect of FH inhibition and DMF treatment, we treated Jurkat cells and PC3 cells with FHi and DMF. In both cell lines, inhibition of FH activity amplified DMF-regulated ULBP2/5/6 surface expression (Fig. 1I, 1J); however, the PC3 cells were unresponsive to FH inhibition alone, indicating cell line–specific regulation.

NKG2D ligands are regulated at several levels, including gene transcription, RNA/protein stabilization, and shedding from the surface (48). We have previously shown that the HDAC inhibitor FR901228 promotes NKG2D ligands by transcriptional activation (20, 22).

To examine whether fumarate regulated the NKG2D ligands at the transcriptional level, we treated Jurkat cells with DMF before doing quantitative PCR analysis. We found that DMF promoted transcriptional activation of ULBP2 with a level almost similar to HDAC inhibitor FR901228 treatment (Fig. 2A). Moreover, DMF upregulated transcription of ULBP5 (Fig. 2B), whereas it was not possible to detect the ULBP6 transcript in response to DMF. The Ab used to label ULBP2/5/6 cannot distinguish between these three homologous ULBP proteins; however, because ULBP6 was not detected in response to DMF treatment, we use the term ULBP2/5 when referring to this staining in the remainder of the study. Notably, DMF did not cause transcriptional activation of MICA/B (Fig. 2C), indicating that fumarate promotes surface expression of ULBP2/5 and MICA/B in Jurkat cells through different mechanisms involving transcriptional and posttranscriptional regulation, respectively.

FIGURE 2.

Fumarate promotes transcriptional activation of ULBP2 and ULBP5. Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 for 4 h. Total RNA was extracted from cells and analyzed by quantitative real-time PCR. Bar graphs show ULBP2 (A), ULBP5 (B), or MICA (C) mRNA expression relative to DMSO-treated cells calculated by the 2−ΔΔCt method. Data are representative of three independent experiments. *p < 0.05, ***p < 0.001.

FIGURE 2.

Fumarate promotes transcriptional activation of ULBP2 and ULBP5. Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 for 4 h. Total RNA was extracted from cells and analyzed by quantitative real-time PCR. Bar graphs show ULBP2 (A), ULBP5 (B), or MICA (C) mRNA expression relative to DMSO-treated cells calculated by the 2−ΔΔCt method. Data are representative of three independent experiments. *p < 0.05, ***p < 0.001.

Close modal

To investigate the functional ability of NKG2D ligands upregulated by fumarate, we examined whether DMF treatment increased overall binding of a soluble form of the native NKG2D receptor. Indeed, the chimeric NKG2D–Fc receptor interacted with NKG2D ligands upregulated by DMF (Fig. 3A), validating that NKG2D ligand/receptor interaction do occur.

FIGURE 3.

ULBP2/5 upregulated by fumarate are functionally active. (A) Jurkat cells were treated with DMSO or 30 μM DMF for 18 h. Cells were stained with soluble NKG2D-Fc and analyzed for NKG2D receptor interaction by flow cytometry. Representative dot plots are shown. (B and C) THP1 cells were treated with DMSO or 100μM DMF for 24 h before being cocultivated with 2B4 or 2B4_NKG2D reporter cells. (B) Increased surface expression of ULBP2/5/6 on DMF-treated THP1 cells was validated before cocultivation. Bar graph shows percentage of positive THP1 cells relative to individual isotype controls. (C) Surface expression of human NKG2D was verified on 2B4_NKG2D reporter cells before cocultivation. Bar graph shows percentage of positive reporter cells relative to individual isotype controls. Representative dot plots of NKG2D expression on 2B4 and 2B4_NKG2D reporter cells. (D) 2B4 or 2B4_NKG2D cells (effector cells) were analyzed for GFP expression (corresponding to NKG2D receptor activation) after cocultivation with THP1 cells (target cells) at the indicated E:T ratio. Graph shows percentage of GFP-positive cells relative to either 2B4 cells or 2B4_NKG2D cells that had not been cocultivated with THP1 cells (1:0 ratio). (E) CD4+-depleted PBLs (effector cells) were activated with IL-15 (10 ng/ml) for 3 d before cocultivated at the indicated E:T ratio for 2 h with THP1 cells (target cells) treated with DMSO, 100 μM DMF, or 20 ng/ml FR901228. Surface expression of NKG2D on CD4+-depleted PBLs is shown as percentage of NKG2D-positive CD4+-depleted PBLs relative to NKG2D receptor blocking. Data are representative of three independent experiments. (F) NK cells (effector cells) were activated with IL-2 (500 U/ml) and IL-15 (10 ng/ml) o/n before being cocultivated at the indicated ratio for 3.5 h with THP1 cells (target cells) treated with DMSO or 100 μM DMF. Surface expression of NKG2D on NK cells is shown as percentage of NKG2D-positive NK cells relative to NKG2D receptor blocking. Data represent two independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 3.

ULBP2/5 upregulated by fumarate are functionally active. (A) Jurkat cells were treated with DMSO or 30 μM DMF for 18 h. Cells were stained with soluble NKG2D-Fc and analyzed for NKG2D receptor interaction by flow cytometry. Representative dot plots are shown. (B and C) THP1 cells were treated with DMSO or 100μM DMF for 24 h before being cocultivated with 2B4 or 2B4_NKG2D reporter cells. (B) Increased surface expression of ULBP2/5/6 on DMF-treated THP1 cells was validated before cocultivation. Bar graph shows percentage of positive THP1 cells relative to individual isotype controls. (C) Surface expression of human NKG2D was verified on 2B4_NKG2D reporter cells before cocultivation. Bar graph shows percentage of positive reporter cells relative to individual isotype controls. Representative dot plots of NKG2D expression on 2B4 and 2B4_NKG2D reporter cells. (D) 2B4 or 2B4_NKG2D cells (effector cells) were analyzed for GFP expression (corresponding to NKG2D receptor activation) after cocultivation with THP1 cells (target cells) at the indicated E:T ratio. Graph shows percentage of GFP-positive cells relative to either 2B4 cells or 2B4_NKG2D cells that had not been cocultivated with THP1 cells (1:0 ratio). (E) CD4+-depleted PBLs (effector cells) were activated with IL-15 (10 ng/ml) for 3 d before cocultivated at the indicated E:T ratio for 2 h with THP1 cells (target cells) treated with DMSO, 100 μM DMF, or 20 ng/ml FR901228. Surface expression of NKG2D on CD4+-depleted PBLs is shown as percentage of NKG2D-positive CD4+-depleted PBLs relative to NKG2D receptor blocking. Data are representative of three independent experiments. (F) NK cells (effector cells) were activated with IL-2 (500 U/ml) and IL-15 (10 ng/ml) o/n before being cocultivated at the indicated ratio for 3.5 h with THP1 cells (target cells) treated with DMSO or 100 μM DMF. Surface expression of NKG2D on NK cells is shown as percentage of NKG2D-positive NK cells relative to NKG2D receptor blocking. Data represent two independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Close modal

To further determine whether ULBP2/5 upregulated by DMF directly activated NKG2D receptor signaling, we used the NKG2D reporter cell line 2B4_NKG2D (49, 50). This reporter cell line expresses GFP in response to NFAT and stably expresses the human NKG2D receptor coupled to DAP10-CD3ζ. The reporter cell line 2B4, which only expresses GFP/NFAT and not NKG2D was included as a control. We used the leukemic monocytic THP1 cell line for this assay because of low basal expression of ULBP2/5. THP1 cells were treated with vehicle or DMF for 24 h before cocultivation with either 2B4 or 2B4_NKG2D reporter cells overnight. As expected, DMF upregulated ULBP2/5 surface expression on THP1 cells (Fig. 3B), and the NKG2D receptor was found only on 2B4_NKG2D cells that were almost 100% positive (Fig. 3C). We found that DMF-treated THP1 cells, in contrast to DMSO-treated THP1 cells, induced a robust signal (30%) through the NKG2D receptor after cocultivation with 2B4_NKG2D cells in an effector/target–dependent manner (Fig. 3D). The activation of the NKG2D receptor was blocked when THP1 cells had been preincubated with NKG2D-Fc, confirming that NKG2D receptor activation was mediated by NKG2D ligands (Supplemental Fig. 2A).

After ligand binding, the NKG2D receptor is internalized upon signaling and cell activation (51). To examine whether DMF-upregulated ULBP2/5 expression resulted in NKG2D down-modulation, we cocultivated DMF-treated THP1 cells with IL-15–activated CD4+-depleted PBLs for 2 h and monitored NKG2D down-modulation. FR901228-treated THP1 cells were included as a positive control for NKG2D down-modulation. Notably, DMF-treated THP1 cells down-modulated the NKG2D receptor from CD4+-depleted PBLs in an effector/target–dependent manner compared with DMSO-treated THP1 cells (Fig. 3E). A similar NKG2D down-modulation was found on purified NK cells activated with IL-2 and IL-15 and cocultivated with DMF-treated THP1 cells (Fig. 3F).

Combined, these data show that ULBP2/5 upregulated by DMF are functionally active, initiate NKG2D receptor signaling, and down-modulate NKG2D surface expression.

Fumarate accumulation caused by FH loss induces oxidative stress as observed in FH-deficient mouse kidney cells with elevated expression of oxidative genes (43). Moreover, oxidative stress and production of ROS associate with NKG2D ligand expression through mechanisms, including metabolic and genotoxic stress, ultimately leading to DNA damage (52, 53). This prompted us to study whether the increased level of fumarate resulted in production of ROS.

We used the fluorogenic reagent MitoSOX to detect mitochondrial ROS (mROS) (superoxide). Jurkat cells were treated with DMF or FR901228 at different time points before they were stained with MitoSOX and analyzed for ROS production. Interestingly, DMF but not FR901228 induced an early production of ROS peaking at 4 h after treatment (Fig. 4A). In contrast, FR901228 resulted in a later production of ROS as observed after 18 h of treatment, whereas DMF-induced ROS production was almost neutralized at this time (Fig. 4B). We tested another fluorogenic reagent, H2DCFDA, which detects cytosolic ROS (hydrogen peroxide). DMF treatment also induced H2DCFDA oxidation, although it was not as potent as MitoSOX oxidation (Supplemental Fig. 3A), suggesting that cytosolic ROS also increases.

FIGURE 4.

The antioxidants NAc and GSH abrogate fumarate-upregulated ULBP2/5 expression. (A) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 for the indicated time. Cells were stained with the fluorogenic reagent MitoSOX to detect mROS (superoxide) before being analyzed by flow cytometry for mROS production. Dot plots show gating strategy of ROS-positive cells and bar graphs show percentage of positive cells within gate. (B) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 for 18 h before being analyzed for ROS production. (C) Jurkat cells were treated with DMSO, 30 μM DMF, 20 ng/ml FR901228, or 10 mM propionate for 18 h. Cells were lysed in RIPA buffer, and p21 protein expression was monitored by Western blot. GAPDH was included as loading control. (D) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 with or without 5 mM NAc for 18 h. Cells were analyzed for ULBP2/5/6 surface expression (left panel). Representative dot plots of Jurkat cells (right panel). (E) Jurkat cells were treated with DMSO, 30 μM DMF or 20 ng/ml FR901228 with or without 100 μM GSH or 100μM GSSG for 18 h and analyzed for ULBP2/5/6 surface expression (left panel). Representative dot plots of Jurkat cells (right panel). (F) Jurkat cells were treated with 50 μM of the indicated compounds for 18 h before being analyzed for surface expression of ULBP2/5/6. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 4.

The antioxidants NAc and GSH abrogate fumarate-upregulated ULBP2/5 expression. (A) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 for the indicated time. Cells were stained with the fluorogenic reagent MitoSOX to detect mROS (superoxide) before being analyzed by flow cytometry for mROS production. Dot plots show gating strategy of ROS-positive cells and bar graphs show percentage of positive cells within gate. (B) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 for 18 h before being analyzed for ROS production. (C) Jurkat cells were treated with DMSO, 30 μM DMF, 20 ng/ml FR901228, or 10 mM propionate for 18 h. Cells were lysed in RIPA buffer, and p21 protein expression was monitored by Western blot. GAPDH was included as loading control. (D) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 with or without 5 mM NAc for 18 h. Cells were analyzed for ULBP2/5/6 surface expression (left panel). Representative dot plots of Jurkat cells (right panel). (E) Jurkat cells were treated with DMSO, 30 μM DMF or 20 ng/ml FR901228 with or without 100 μM GSH or 100μM GSSG for 18 h and analyzed for ULBP2/5/6 surface expression (left panel). Representative dot plots of Jurkat cells (right panel). (F) Jurkat cells were treated with 50 μM of the indicated compounds for 18 h before being analyzed for surface expression of ULBP2/5/6. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Close modal

To address whether fumarate resulted in genotoxic stress, Jurkat cells were treated with DMF and analyzed for p21 expression by Western blot. p21 is a target of the tumor suppressor gene p53 and responsible for linking DNA damage to cell cycle arrest. We included the HDAC inhibitor FR901228 and the SCFA propionate that are well-known inducers of NKG2D ligands (17, 20). No induction of p21 expression was observed after DMF treatment (Fig. 4C), which was in contrast to HDAC inhibitor and SCFA treatment that induced a robust p21 induction. Together, our data suggest that DMF induces metabolically derived ROS production without causing p21-mediated DNA damage response.

To further study the impact of ROS production, we investigated the effect of the antioxidant NAc on DMF-upregulated ULBP2/5 expression. Jurkat cells were treated with vehicle or NAc before DMF treatment. Interestingly, NAc abrogated ULBP2/5 surface expression upregulated by DMF (Fig. 4D) and reduced MICA/B surface expression (Supplemental Fig. 3B). In line with our previous finding (54), ULBP2/5 and MICA/B surface expression upregulated by HDAC inhibitor treatment were unaffected by NAc treatment. Combined with the differential p21 DNA damage activation, this suggests that DMF and HDAC-inhibitors upregulate NKG2D ligands through distinct pathways.

We next wanted to specify which antioxidant system that regulated ULBP2/5 was promoted by DMF. NAc is a biological precursor of GSH. GSH is synthesized in many cell types from the amino acids glutamine, cysteine, and glycine and functions as a major antioxidant to neutralize cellular stress caused by ROS (30). More specifically, GSH works by reducing target molecules in a process that oxidizes GSH to GSSG. To investigate the effect of GSH and GSSG, we treated Jurkat cells with GSH or GSSG before treatment with DMF. Similar to NAc, GSH completely abrogated ULBP2/5 surface expression upregulated by DMF but not by the HDAC-inhibitor FR901228 (Fig. 4E). Oxidized GSSG did not inhibit ULBP2/5 surface expression, indicating that the reducing capacity of GSH is essential to neutralize DMF-upregulated ULBP2/5 response. GSH treatment also reduced DMF-upregulated MICA/B surface expression (Supplemental Fig. 3C).

Intracellular fumarate causes acute toxicity that is neutralized by the direct binding of GSH to fumarate in a process producing GSF (9, 43). To validate that the reducing capacity of GSH is neutralizing ULBP2/5 upregulated by DMF, we complexed DMF with GSH (generating GSF) prior to treatment of Jurkat cells (9). As expected, GSF did not upregulate ULBP2/5 (Fig. 4F) or MICA/B surface expression (Supplemental Fig. 3D).

The antioxidant capacity of GSH is dependent on its regeneration from GSSG. The enzyme GR catalyzes the reduction of GSSG into GSH, thereby restoring the antioxidant capacity (30). As our data indicate a key role of GSH in regulation of ULBP2/5 by fumarate, we investigated whether GSH recycling affected ULBP2/5 expression.

To study this, we used the irreversible GR inhibitor BCNU, resulting in defective cellular GSH activity (55). Jurkat cells were treated with BCNU and monitored for ULBP2/5 surface expression. Inhibition of GR activity resulted in increased ULBP2/5 surface expression (Fig. 5A) but failed to upregulate MICA/B surface expression (Supplemental Fig. 3E). This is in line with our current data showing that ULBP2/5 compared with MICA/B is more potently upregulated after DMF treatment. Recycling of GSSG to GSH by GR happens at the expense of NADPH donating the reducing electrons. The main cellular pathway producing NADPH is the pentose phosphate pathway (PPP) that diverts from proximal glycolysis (56). To validate the essential role of GR activity for maintaining homeostatic expression of ULBP2/5, we treated Jurkat cells with the PPP inhibitor 6-AN and analyzed ULBP2/5 expression. Similar to GR inhibition, depletion of NADPH and thus inhibition of GSH recycling promoted surface expression of ULBP2/5 (Fig. 5B) but not MICA/B (Supplemental Fig. 3F).

FIGURE 5.

Blockade of GSH recycling upregulates surface expression of ULBP2/5. (A) Jurkat cells were treated with DMSO or the indicated concentrations of the GR inhibitor BCNU for 18 h. Cells were analyzed by flow cytometry for surface expression of ULBP2/5/6. Representative dot plots are shown. (B) Jurkat cells were treated with DMSO or 100μM PPP inhibitor 6-AN for 48 h before being analyzed for ULBP2/5/6 surface expression. Representative dot plots are shown. (C) Representative dot plots of Jurkat cell treated with DMSO, 80 μM BCNU, or 100 μM 6-AN for 18 h and analyzed for mROS production using MitoSOX and cytosolic ROS production using H2DCFDA. (D) Jurkat cells were treated with DMSO or 80 μM BCNU with or without 5 mM NAc or 100μM GSH for 18 h. (E) Dot plots show Jurkat cells in (D). (F) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 with or without 40μM BCNU before being analyzed for ULBP2/5/6 surface expression. (G) Schematic overview of accumulated fumarate that interacts with GSH recycling, ultimately leading to oxidative stress and ULBP2/5/6 surface expression. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 5.

Blockade of GSH recycling upregulates surface expression of ULBP2/5. (A) Jurkat cells were treated with DMSO or the indicated concentrations of the GR inhibitor BCNU for 18 h. Cells were analyzed by flow cytometry for surface expression of ULBP2/5/6. Representative dot plots are shown. (B) Jurkat cells were treated with DMSO or 100μM PPP inhibitor 6-AN for 48 h before being analyzed for ULBP2/5/6 surface expression. Representative dot plots are shown. (C) Representative dot plots of Jurkat cell treated with DMSO, 80 μM BCNU, or 100 μM 6-AN for 18 h and analyzed for mROS production using MitoSOX and cytosolic ROS production using H2DCFDA. (D) Jurkat cells were treated with DMSO or 80 μM BCNU with or without 5 mM NAc or 100μM GSH for 18 h. (E) Dot plots show Jurkat cells in (D). (F) Jurkat cells were treated with DMSO, 30 μM DMF, or 20 ng/ml FR901228 with or without 40μM BCNU before being analyzed for ULBP2/5/6 surface expression. (G) Schematic overview of accumulated fumarate that interacts with GSH recycling, ultimately leading to oxidative stress and ULBP2/5/6 surface expression. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Close modal

To study whether inhibition of GSH recycling resulted in increased production of ROS, we treated Jurkat cells with BCNU or 6-AN and monitored ROS production by MitoSOX and H2DCFDA. BCNU but not 6-AN treatment induced ROS production, indicating that direct inhibition of GSH recycling more efficiently induces oxidative stress (Fig. 5C). Similar to DMF treatment, we found that the antioxidants NAc and GSH reduced ULBP2/5 surface expression upregulated by BCNU, although not to the extent observed after DMF treatment (Fig. 5D, 5E). We tested the impact of GR inhibition on DMF or FR901228-upregulated ULBP2/5 surface expression. BCNU amplified ULBP2/5 expression after DMF and FR901228 treatment compared with cells treated with individual stimuli (Fig. 5F).

Our data indicate that fumarate upregulates ULBP2/5 surface expression by scavenging cells of GSH-mediated antioxidant capacity (Fig. 5G). We propose that GSH neutralizes fumarate to GSF. In this way, fumarate uses the intracellular pool of GSH, eventually exhausting cells from antioxidant capacity. Ultimately, the elevated oxidative stress translates into ULBP2/5 surface expression. A similar ULBP2/5 response is promoted in the absence of fumarate by depleting cells of GSH-mediated antioxidant capacity (BCNU treatment), indicating that impairment of GSH recycling is sufficient to upregulate ULBP2/5 surface expression.

The citric acid cycle enzyme FH is a tumor suppressor gene associated with malignant renal cancer. Loss of FH activity is causally involved in development of HLRCC, which is characterized by citric acid cycle breakage, accumulation of fumarate, and high ROS production (911). We therefore addressed NKG2D ligand expression in cells lacking FH activity. To study this, we obtained the FH-deficient UOK262 renal cancer cell line derived from a patient with HLRCC (11, 57). UOK262 cells expressed a low basal level of ULBP2/5 and MICA/B (Fig. 6A). Moreover, DMF did not regulate surface expression of ULBP2/5 or MICA/B (Fig. 6A–C), suggesting that chronic high fumarate level has hampered the induction of ULBP2/5 in these cells.

FIGURE 6.

Fumarate does not regulate ULBP2/5 on FH-deficient renal cancer cells. (A) Representative dot plots of UOK262 cells. Left, Basal surface expression of ULBP2/5/6 and MICA/B relative to corresponding isotype controls. Right, Upregulated surface expression of ULBP2/5/6 and MICA/B relative to DMSO-treated control cells. (B and C) UOK262 cells were treated with DMSO or 60μM DMF for 18 h before being analyzed by flow cytometry for surface expression of ULBP2/5/6 or MICA/B. FR901228 (10 ng/ml) and propionate (10 mM) were included as positive controls for NKG2D ligand induction. (D) Representative dot plots of NCCFH1 cells. Left, Basal surface expression of ULBP2/5/6 and MICA/B relative to corresponding isotype controls. Right, Upregulated surface expression of ULBP2/5/6 and MICA/B relative to DMSO-treated control cells. (E and F) NCCFH1 cells were treated with DMSO, 60 μM DMF, 20 ng/ml FR901228, or 10 mM propionate for 18 h before being analyzed by flow cytometry for surface expression of ULBP2/5/6 or MICA/B. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ****p < 0.0001.

FIGURE 6.

Fumarate does not regulate ULBP2/5 on FH-deficient renal cancer cells. (A) Representative dot plots of UOK262 cells. Left, Basal surface expression of ULBP2/5/6 and MICA/B relative to corresponding isotype controls. Right, Upregulated surface expression of ULBP2/5/6 and MICA/B relative to DMSO-treated control cells. (B and C) UOK262 cells were treated with DMSO or 60μM DMF for 18 h before being analyzed by flow cytometry for surface expression of ULBP2/5/6 or MICA/B. FR901228 (10 ng/ml) and propionate (10 mM) were included as positive controls for NKG2D ligand induction. (D) Representative dot plots of NCCFH1 cells. Left, Basal surface expression of ULBP2/5/6 and MICA/B relative to corresponding isotype controls. Right, Upregulated surface expression of ULBP2/5/6 and MICA/B relative to DMSO-treated control cells. (E and F) NCCFH1 cells were treated with DMSO, 60 μM DMF, 20 ng/ml FR901228, or 10 mM propionate for 18 h before being analyzed by flow cytometry for surface expression of ULBP2/5/6 or MICA/B. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ****p < 0.0001.

Close modal

To validate that UOK262 cells were able to induce NKG2D ligands after cellular stress, we included the HDAC-inhibitor FR901228 and the SCFA propionate. Both of these treatments upregulated MICA/B surface expression (Fig. 6A, 6C) but failed to induce a substantial ULBP2/5 response (Fig. 6A, 6B).

FH-deficiency associates with increased stabilization of hypoxia-inducible factors (HIFs) and activation of Nrf2 signaling (12, 58, 59). These cellular adaptations are likely essential for FH-deficient cancer cells to survive fumarate-induced ROS accumulation and sustain proliferation. We speculated that DMF failed to regulate ULBP2/5 in FH-deficient cancer cells because of high basal activation of antioxidant response that rendered the cells less susceptible to ROS-mediated stress. However, when targeting the major antioxidant pathways in UOK262 cells using BCNU or 6-AN, we observed no induction of ULBP2/5 expression on the transcriptional or surface level (Supplemental Fig. 4A, 4B), suggesting that chronic fumarate accumulation causes a more general deficiency in ULBP2/5 regulation. To examine whether ULBP2 could traffic to the cell surface in UOK262 cells, we transfected the cells with a myc-tagged GFP–ULBP2 construct and found surface expression of ULBP2, indicating that FH-deficient cells do not inhibit the transport of ULBP2 to the surface (Supplemental Fig. 4C).

We tested another FH-deficient cell line, NCCFH1, established from a patient with hereditary papillary renal cell carcinoma type 2 (60). NCCFH1 cells expressed a detectable basal level of ULBP2/5 and MICA/B (Fig. 6D). Similar to UOK262 cells, NCCFH1 cells did not express ULBP2/5 (Fig. 6D, 6E) and MICA/B (Fig. 6D, 6F) in response to DMF treatment. However, they did express more substantial ULBP2/5 after treatment with FR901228 and propionate (Fig. 6D), indicating that NCCFH1 cells hamper ULBP2/5 expression by a different mechanism than UOK262, allowing differential expression. ULBP2/5 expression was not induced on the transcriptional (Supplemental Fig. 4D) or surface level (Supplemental Fig. 4E) by targeting antioxidant pathways in NCCFH1 cells using BCNU.

Together, these data indicate that chronic accumulation of fumarate as observed in FH-deficient cancer cells renders cells unresponsive to DMF treatment, suggesting that they decouple fumarate accumulation from ULBP2/5 surface expression. This could possibly serve as an immune-evasive strategy that enables FH-deficient cancer cells to avoid immune recognition despite buildup of fumarate.

We were interested to identify whether fumarate accumulation itself or a product prior to fumarate resulted in increased ULBP2/5 surface expression. SDH is a citric acid cycle enzyme that converts succinate to fumarate (61).

To study whether ULBP2/5 expression was a direct effect of fumarate accumulation, we blocked the activity of SDH prior to DMF treatment using malonate, a competitive SDHi. Interestingly, inhibition of SDH amplified DMF-upregulated ULBP2/5 surface expression in a concentration-dependent manner (Fig. 7A), suggesting that ULBP2/5 is not only linked to fumarate accumulation but that buildup of the immediate upstream succinate also promotes ULBP2/5 surface expression. To study this, we treated Jurkat cells with CP-Succ, which is specifically designed to penetrate cells and support respiration in complex I deficiency (62). CP-Succ activated gene expression of ULBP2 in Jurkat cells (Fig. 7B) as well as ULBP2/5 surface expression on Jurkat cells (Fig. 7C, 7D), PC3 cells (Fig. 7E), and THP1 cells (Fig. 7F). When treating Jurkat cells with SDHi prior to CP-Succ treatment, SDH inhibition amplified CP-Succ–upregulated ULBP2/5 surface expression (Fig. 7G). In line with this, higher concentrations of SDHi were sufficient to upregulate ULBP2/5 surface expression in the absence of CP-Succ treatment (Fig. 7H), suggesting that accumulation of succinate translates into ULBP2/5 expression.

FIGURE 7.

Both fumarate and succinate accumulation upregulate ULBP2/5 but through different mechanisms. (A) Jurkat cells were treated with DMSO or 30μM DMF with or without the indicated concentrations of SDHi for 18 h and analyzed for ULBP2/5/6 expression. (B) Jurkat cells were treated with DMSO or 50–70 μM CP-Succ for 4 h before total RNA was extracted and analyzed using quantitative real-time PCR. Bar graph shows ULBP2 expression normalized to RPLP0 and relative to DMSO-treated cells. (C) Representative dot plots of cells in (D). (D) Jurkat cells were treated with the indicated concentrations of the CP-Succ for 18 h before being analyzed for surface expression of ULBP2/5/6. Surface expression of ULBP2/5/6 on PC3 cells (E) treated with DMSO or 80 μM CP-Succ for 18 h or on THP1 cells (F) treated with DMSO or 50 μM CP-Succ for 24 h. (G) Jurkat cells were treated with DMSO or 50 μM CP-Succ with or without the indicated concentrations of SDHi for 18 h before being analyzed for ULBP2/5/6 expression. (H) Jurkat cells were treated with the indicated concentrations of SDHi for 18 h and analyzed for ULBP2/5/6 expression. (I) NCCFH1 cells were treated with DMSO, 60 μM DMF, or 60 μM CP-Succ for 18 h and analyzed for ULBP2/5/6 expression. (J) UOK262 cells were treated with DMSO, 60 μM DMF, or 60 μM CP-Succ for 18 h before being analyzed for ULBP2/5/6 expression. (K) Representative dot plots of Jurkat cells treated with DMSO or 70 μM CP-Succ for 18 h and analyzed with MitoSOX and H2DCFDA to detect ROS production. (L) Jurkat cells were treated with DMSO or 70 μM CP-Succ with or without 5 mM NAc for 18 h and analyzed for ULBP2/5/6 expression. (M) Jurkat cells were treated with DMSO or 70 μM CP-Succ with or without 100 μM GSH or 100 μM GSSG for 18 h before being analyzed for ULBP2/5/6 expression. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

FIGURE 7.

Both fumarate and succinate accumulation upregulate ULBP2/5 but through different mechanisms. (A) Jurkat cells were treated with DMSO or 30μM DMF with or without the indicated concentrations of SDHi for 18 h and analyzed for ULBP2/5/6 expression. (B) Jurkat cells were treated with DMSO or 50–70 μM CP-Succ for 4 h before total RNA was extracted and analyzed using quantitative real-time PCR. Bar graph shows ULBP2 expression normalized to RPLP0 and relative to DMSO-treated cells. (C) Representative dot plots of cells in (D). (D) Jurkat cells were treated with the indicated concentrations of the CP-Succ for 18 h before being analyzed for surface expression of ULBP2/5/6. Surface expression of ULBP2/5/6 on PC3 cells (E) treated with DMSO or 80 μM CP-Succ for 18 h or on THP1 cells (F) treated with DMSO or 50 μM CP-Succ for 24 h. (G) Jurkat cells were treated with DMSO or 50 μM CP-Succ with or without the indicated concentrations of SDHi for 18 h before being analyzed for ULBP2/5/6 expression. (H) Jurkat cells were treated with the indicated concentrations of SDHi for 18 h and analyzed for ULBP2/5/6 expression. (I) NCCFH1 cells were treated with DMSO, 60 μM DMF, or 60 μM CP-Succ for 18 h and analyzed for ULBP2/5/6 expression. (J) UOK262 cells were treated with DMSO, 60 μM DMF, or 60 μM CP-Succ for 18 h before being analyzed for ULBP2/5/6 expression. (K) Representative dot plots of Jurkat cells treated with DMSO or 70 μM CP-Succ for 18 h and analyzed with MitoSOX and H2DCFDA to detect ROS production. (L) Jurkat cells were treated with DMSO or 70 μM CP-Succ with or without 5 mM NAc for 18 h and analyzed for ULBP2/5/6 expression. (M) Jurkat cells were treated with DMSO or 70 μM CP-Succ with or without 100 μM GSH or 100 μM GSSG for 18 h before being analyzed for ULBP2/5/6 expression. Data are representative of three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Close modal

To further examine the mechanisms used by fumarate and succinate, we tested the effect of succinate accumulation in FH-deficient UOK262 and NCCFH1 cell lines. In contrast to DMF, which did not induce ULBP2/5 surface expression on the FH-deficient cell lines, CP-Succ upregulated ULBP2/5 on NCCFH1 cells (Fig. 7I). This indicates that NCCFH1 cells obstruct ULBP2/5 expression by fumarate but not succinate, suggesting that fumarate and succinate use different pathways to upregulate ULBP2/5. CP-Succ did not induce ULBP2/5 on UOK262 cells (Fig. 7J), confirming that these cells have a more general deficiency in ULBP2/5 expression.

Similar to fumarate, accumulation of succinate associates with increased ROS production and activation of HIF-1 (63, 64). In line with this, ROS production was increased after CP-Succ treatment as monitored by MitoSOX and H2DCFDA oxidation (Fig. 7K). Moreover, ULBP2/5 surface expression upregulated by CP-Succ was abrogated by NAc (Fig. 7L) and GSH (Fig. 7M), indicating that oxidative signaling is also involved in ULBP2/5 expression mediated by succinate.

Altogether, our data suggest that both fumarate and succinate upregulate ULBP2/5, likely through different mechanisms. This is, however, only clearly observed in FH-deficient NCCFH1 cells. Depending on the direction of the citric acid cycle, both fumarate and succinate are likely to accumulate together and share similar dependence on oxidative signaling, leading to ULBP2/5 surface expression.

Studies have identified fumarate as an oncometabolite that modifies cell metabolism, activates stress pathways, and induces aberrant HIF signaling (8, 9, 12). In this study, we show that acute inhibition of FH activity leading to fumarate accumulation results in surface expression of the NKG2D ligands ULBP2/5. A functional ULBP2/5 expression was also observed after artificial intracellular fumarate accumulation, strongly indicating that intracellular accumulation of fumarate regulates NKG2D-mediated immune sensing.

When examining the functionality of DMF-upregulated NKG2D ligands in a NKG2D down-modulation assay, we found that NKG2D was down-modulated on CD4+-depleted PBLs and on purified NK cells. NKG2D down-modulation has been associated with NK cell exhaustion during chronic stimulation (65). NK cell exhaustion was likely not involved in the NKG2D down-modulation we observe after DMF-upregulated NKG2D ligand engagement, as this could be specifically blocked by inclusion of NKG2D-Fc. In addition, NKG2D down-modulation was monitored after a few hours of cocultivation, which is likely not enough time for NK cell exhaustion.

The antioxidant GSH maintains redox homeostasis and prevents cellular damage caused by ROS. Fumarate can react with GSH to form the cancer-associated metabolite GSF that serves as an alternative substrate to GR, ultimately exhausting cells of functional GSH and NADPH (9). Similar to fumarate accumulation, blockade of functional GSH activity resulted in ULBP2/5 surface expression, whereas treatment with the GSH precursor NAc as well as GSH abrogated fumarate-regulated ULBP2/5 surface expression, suggesting that accumulation of fumarate upregulates ULBP2/5 expression by exhausting cells of functional GSH.

Highly proliferating cancer cells and activated T cells generate ATP by Warburg metabolism, which facilitates rapid ATP production through glycolysis rather than through oxidative phosphorylation (66). Warburg metabolism also enables high PPP activity, with production of NADPH used in lipid synthesis and GSH recycling (56). We show that blockade of GSH activity by inhibiting NADPH production or GSH recycling also resulted in ULBP2/5 surface expression. A destabilized GSH system will, by itself, thus lead to ULBP2/5 expression, further supporting that fumarate causes ULBP2/5 expression by exhausting GSH function.

Many cancer cells upregulate genes specifically involved in GSH and NADPH synthesis; examples include the rate-limiting enzyme GCL involved in de novo GSH synthesis and NADP-dependent isocitrate dehydrogenase (IDH) 1 (67, 68). IDH mutations are previously linked to down-modulation of NKG2D ligand expression in glioma cells (69). Increased synthesis of GSH and NADPH enables cancer cells to cope with higher concentrations of ROS and resist chemotherapeutic drugs such as cisplatin and doxorubicin (70, 71). GSH activity in cancer cells is therefore an attractive target in cancer therapy. To our knowledge, NKG2D/NKG2D ligands have not been investigated in this respect. Our data indicate that depletion of GSH antioxidant capacity through fumarate treatment results in NKG2D ligand expression, thereby adding to the therapeutic potential of targeting GSH activity.

We found that artificial accumulation of intracellular fumarate also resulted in surface expression of the NKG2D ligands MICA/B. Compared with ULBP2/5, the induction was weaker and did not relate to increased transcription, suggesting that fumarate upregulates NKG2D ligands through several mechanisms. The functional outcome of NKG2D receptor engagement is governed by the combined expression of NKG2D ligands. By regulating NKG2D ligands through different mechanisms, fumarate may enable a more versatile activation of the NKG2D receptor.

Intriguingly, our study showed that FH-deficient cells with chronic accumulation of fumarate had low basal expression of ULBP2/5 that was not increased after DMF treatment. FH-deficient cells accumulate fumarate that complexes with GSH, thus depleting cells of NADPH and ultimately elevating ROS level and signaling (9). To control the high ROS level, FH-deficient cells increase the antioxidant transcription factor Nrf2, an adaptive response that is partly mediated by the direct binding of fumarate to KEAP1 (59). However, targeting GSH-mediated antioxidant response in FH-deficient cells did not induce ULBP2/5 expression. Our data thus indicate that the continuous high level of fumarate decouples from ULBP2/5 expression, potentially serving as a mechanism to escape immune recognition despite buildup of fumarate.

We found that the citric acid cycle metabolite succinate also resulted in ULBP2/5 surface expression on several cancer cells. Interestingly, our data showed that succinate, in contrast to fumarate, upregulated ULBP2/5 on FH-deficient NCCFH1 cells, suggesting that the metabolites regulate ULBP2/5 expression through different mechanisms. We speculate that succinate induces a more direct ROS production through SDH/complex II activity (72). Fumarate, in contrast, mainly causes ROS production because of reduced GSH antioxidant activity. FH-deficient NCCFH1 cells have acquired resistance toward the diminished GSH activity, but they are still sensitive to the direct increase in ROS. Because succinate is immediately upstream of fumarate, an increased fumarate level because of a citric acid cycle break will associate with an increased succinate level. In fact, some cancer cells reverse the reaction of SDH, converting fumarate to succinate, thus allowing regeneration of NAD+ and FAD+ under sparse oxygen supply (72). However, no matter the interplay, accumulation of fumarate, succinate, or the combination will lead to increased ULBP2/5 surface expression.

Succinate can bind to the G-protein–coupled receptor GPR91 (known as SUCNR1) and thereby amplify production of proinflammatory cytokines under conditions of LPS-treatment and arthritis (7375). However, because only a cell-permeable form of succinate and not succinate itself upregulates NKG2D ligands, it is unlikely that SUCNR1 is causally involved in the response we observe (20).

In this study, we used DMF to artificially induce fumarate accumulation. DMF is used as a therapeutic drug to treat the autoimmune disorders multiple sclerosis and psoriasis, but it has also shown potential as an anticancer drug through its antiproliferative and proapoptotic effects in melanoma cells (44). Our data strongly suggest that DMF treatment upregulates ULBP2/5 surface expression by scavenging GSH-mediated antioxidant response, further highlighting DMF as a promising anticancer drug.

We thank Prof. Carsten Geisler (Department of Immunology and Microbiology, University of Copenhagen, Denmark) for providing Jurkat TAg9 T cells, Dr. W. Marston Linehan (Center for Cancer Research, National Cancer Institute, Bethesda, MD) for providing FH-deficient UOK262 cells, Dr. Min Han-Tan and Dr. Bin Tean The (Institute of Bioengineering and Nanotechnology, The Nanos, Singapore and Laboratory of Cancer Epigenome, National Cancer Center Singapore) for providing FH-deficient NCCFH1 cells, and Dr. Chiwen Chang (Department of Pathology, University of Cambridge, Cambridge, U.K.) for providing 2B4 and 2B4_NKG2D cells. Also, we thank to Anni Mehlsen (Department of Veterinary and Animal Sciences, University of Copenhagen, Copenhagen, Denmark) for technical assistance.

This work was supported by the Danish Council for Independent Research (DFF-6111-00499) and the Novo Nordisk Foundation (NNF15CC0018346).

The online version of this article contains supplemental material.

Abbreviations used in this article:

6-AN

6-aminonicotinamide

BCNU

1,3-bis(2-chloroethyl)-1-nitrosourea

CP-Succ

cell-permeable succinate prodrug

DMF

dimethyl fumarate

FH

fumarate hydratase

FHi

FH-IN-1

GR

GSH reductase

GSF

succinated GSH

GSH

glutathione

GSSG

disulfide-oxidized l-GSH

HDAC

histone deacetylase

H2DCFDA

2′7′-dichlorodihydrofluorescein diacetate

HIF

hypoxia-inducible factor

HLRCC

hereditary leiomyomatosis renal cell carcinoma

MIC

MHC class I polypeptide-related sequence

mROS

mitochondrial ROS

NAc

N-acetylcysteine

NKG2D

NK group 2, member D

o/n

overnight

PPP

pentose phosphate pathway

ROS

reactive oxygen species

RPLP0

ribosomal protein, large P0

SCFA

short chain fatty acid

SDH

succinate dehydrogenase

SDHi

SDH inhibitor.

1
Cairns
,
R. A.
,
I. S.
Harris
,
T. W.
Mak
.
2011
.
Regulation of cancer cell metabolism.
Nat. Rev. Cancer
11
:
85
95
.
2
Pavlova
,
N. N.
,
C. B.
Thompson
.
2016
.
The emerging hallmarks of cancer metabolism.
Cell Metab.
23
:
27
47
.
3
Birsoy
,
K.
,
T.
Wang
,
W. W.
Chen
,
E.
Freinkman
,
M.
Abu-Remaileh
,
D. M.
Sabatini
.
2015
.
An essential role of the mitochondrial Electron transport chain in cell proliferation is to enable aspartate synthesis.
Cell
162
:
540
551
.
4
DeBerardinis
,
R. J.
,
A.
Mancuso
,
E.
Daikhin
,
I.
Nissim
,
M.
Yudkoff
,
S.
Wehrli
,
C. B.
Thompson
.
2007
.
Beyond aerobic glycolysis: transformed cells can engage in glutamine metabolism that exceeds the requirement for protein and nucleotide synthesis.
Proc. Natl. Acad. Sci. USA
104
:
19345
19350
.
5
Ryan
,
D. G.
,
M. P.
Murphy
,
C.
Frezza
,
H. A.
Prag
,
E. T.
Chouchani
,
L. A.
O’Neill
,
E. L.
Mills
.
2019
.
Coupling Krebs cycle metabolites to signalling in immunity and cancer.
Nat Metab
1
:
16
33
.
6
Tannahill
,
G. M.
,
A. M.
Curtis
,
J.
Adamik
,
E. M.
Palsson-McDermott
,
A. F.
McGettrick
,
G.
Goel
,
C.
Frezza
,
N. J.
Bernard
,
B.
Kelly
,
N. H.
Foley
, et al
.
2013
.
Succinate is an inflammatory signal that induces IL-1β through HIF-1α.
Nature
496
:
238
242
.
7
Arts
,
R. J.
,
B.
Novakovic
,
R.
Ter Horst
,
A.
Carvalho
,
S.
Bekkering
,
E.
Lachmandas
,
F.
Rodrigues
,
R.
Silvestre
,
S. C.
Cheng
,
S. Y.
Wang
, et al
.
2016
.
Glutaminolysis and fumarate accumulation integrate immunometabolic and epigenetic programs in trained immunity.
Cell Metab.
24
:
807
819
.
8
Yang
,
M.
,
T.
Soga
,
P. J.
Pollard
,
J.
Adam
.
2012
.
The emerging role of fumarate as an oncometabolite.
Front. Oncol.
2
:
85
.
9
Sullivan
,
L. B.
,
E.
Martinez-Garcia
,
H.
Nguyen
,
A. R.
Mullen
,
E.
Dufour
,
S.
Sudarshan
,
J. D.
Licht
,
R. J.
Deberardinis
,
N. S.
Chandel
.
2013
.
The proto-oncometabolite fumarate binds glutathione to amplify ROS-dependent signaling. [Published erratum appears in Mol. Cell 51: 273.]
Mol. Cell
51
:
236
248
.
10
Tomlinson
,
I. P.
,
N. A.
Alam
,
A. J.
Rowan
,
E.
Barclay
,
E. E.
Jaeger
,
D.
Kelsell
,
I.
Leigh
,
P.
Gorman
,
H.
Lamlum
,
S.
Rahman
, et al
Multiple Leiomyoma Consortium
.
2002
.
Germline mutations in FH predispose to dominantly inherited uterine fibroids, skin leiomyomata and papillary renal cell cancer.
Nat. Genet.
30
:
406
410
.
11
Yang
,
Y.
,
V. A.
Valera
,
H. M.
Padilla-Nash
,
C.
Sourbier
,
C. D.
Vocke
,
M. A.
Vira
,
M. S.
Abu-Asab
,
G.
Bratslavsky
,
M.
Tsokos
,
M. J.
Merino
, et al
.
2010
.
UOK 262 cell line, fumarate hydratase deficient (FH-/FH-) hereditary leiomyomatosis renal cell carcinoma: in vitro and in vivo model of an aberrant energy metabolic pathway in human cancer.
Cancer Genet. Cytogenet.
196
:
45
55
.
12
Pollard
,
P. J.
,
J. J.
Brière
,
N. A.
Alam
,
J.
Barwell
,
E.
Barclay
,
N. C.
Wortham
,
T.
Hunt
,
M.
Mitchell
,
S.
Olpin
,
S. J.
Moat
, et al
.
2005
.
Accumulation of Krebs cycle intermediates and over-expression of HIF1alpha in tumours which result from germline FH and SDH mutations.
Hum. Mol. Genet.
14
:
2231
2239
.
13
Houchins
,
J. P.
,
T.
Yabe
,
C.
McSherry
,
F. H.
Bach
.
1991
.
DNA sequence analysis of NKG2, a family of related cDNA clones encoding type II integral membrane proteins on human natural killer cells.
J. Exp. Med.
173
:
1017
1020
.
14
Champsaur
,
M.
,
L. L.
Lanier
.
2010
.
Effect of NKG2D ligand expression on host immune responses.
Immunol. Rev.
235
:
267
285
.
15
Carapito
,
R.
,
S.
Bahram
.
2015
.
Genetics, genomics, and evolutionary biology of NKG2D ligands.
Immunol. Rev.
267
:
88
116
.
16
Lanier
,
L. L.
2015
.
NKG2D receptor and its ligands in host defense.
Cancer Immunol. Res.
3
:
575
582
.
17
Skov
,
S.
,
M. T.
Pedersen
,
L.
Andresen
,
P. T.
Straten
,
A.
Woetmann
,
N.
Odum
.
2005
.
Cancer cells become susceptible to natural killer cell killing after exposure to histone deacetylase inhibitors due to glycogen synthase kinase-3-dependent expression of MHC class I-related chain A and B.
Cancer Res.
65
:
11136
11145
.
18
Venkataraman
,
G. M.
,
D.
Suciu
,
V.
Groh
,
J. M.
Boss
,
T.
Spies
. (
2007
)
Promoter region architecture and transcriptional regulation of the genes for the MHC class I-related chain A and B ligands of NKG2D.
J. Immunol.
178
:
961
969
.
19
Gasser
,
S.
,
S.
Orsulic
,
E. J.
Brown
,
D. H.
Raulet
.
2005
.
The DNA damage pathway regulates innate immune system ligands of the NKG2D receptor.
Nature
436
:
1186
1190
.
20
Andresen
,
L.
,
K. A.
Hansen
,
H.
Jensen
,
S. F.
Pedersen
,
P.
Stougaard
,
H. R.
Hansen
,
J.
Jurlander
,
S.
Skov
. (
2009
)
Propionic acid secreted from propionibacteria induces NKG2D ligand expression on human-activated T lymphocytes and cancer cells.
J. Immunol.
183
:
897
906
.
21
McCarthy
,
M. T.
,
G.
Moncayo
,
T. K.
Hiron
,
N. A.
Jakobsen
,
A.
Valli
,
T.
Soga
,
J.
Adam
,
C. A.
O’Callaghan
.
2018
.
Purine nucleotide metabolism regulates expression of the human immune ligand MICA.
J. Biol. Chem.
293
:
3913
3924
.
22
Uhlenbrock
,
F.
,
M.
Hagemann-Jensen
,
S.
Kehlet
,
L.
Andresen
,
S.
Pastorekova
,
S.
Skov
. (
2014
)
The NKG2D ligand ULBP2 is specifically regulated through an invariant chain-dependent endosomal pathway.
J. Immunol.
193
:
1654
1665
.
23
Mellergaard
,
M.
,
S. L.
Skovbakke
,
C. L.
Schneider
,
F.
Lauridsen
,
L.
Andresen
,
H.
Jensen
,
S.
Skov
.
2014
.
N-glycosylation of asparagine 8 regulates surface expression of major histocompatibility complex class I chain-related protein A (MICA) alleles dependent on threonine 24.
J. Biol. Chem.
289
:
20078
20091
.
24
Andresen
,
L.
,
S. L.
Skovbakke
,
G.
Persson
,
M.
Hagemann-Jensen
,
K. A.
Hansen
,
H.
Jensen
,
S.
Skov
. (
2012
)
2-deoxy D-glucose prevents cell surface expression of NKG2D ligands through inhibition of N-linked glycosylation.
J. Immunol.
188
:
1847
1855
.
25
Groh
,
V.
,
J.
Wu
,
C.
Yee
,
T.
Spies
.
2002
.
Tumour-derived soluble MIC ligands impair expression of NKG2D and T-cell activation.
Nature
419
:
734
738
.
26
Raffaghello
,
L.
,
I.
Prigione
,
I.
Airoldi
,
M.
Camoriano
,
I.
Levreri
,
C.
Gambini
,
D.
Pende
,
A.
Steinle
,
S.
Ferrone
,
V.
Pistoia
.
2004
.
Downregulation and/or release of NKG2D ligands as immune evasion strategy of human neuroblastoma.
Neoplasia
6
:
558
568
.
27
Paschen
,
A.
,
A.
Sucker
,
B.
Hill
,
I.
Moll
,
M.
Zapatka
,
X. D.
Nguyen
,
G. C.
Sim
,
I.
Gutmann
,
J.
Hassel
,
J. C.
Becker
, et al
.
2009
.
Differential clinical significance of individual NKG2D ligands in melanoma: soluble ULBP2 as an indicator of poor prognosis superior to S100B.
Clin. Cancer Res.
15
:
5208
5215
.
28
Dunn
,
C.
,
N. J.
Chalupny
,
C. L.
Sutherland
,
S.
Dosch
,
P. V.
Sivakumar
,
D. C.
Johnson
,
D.
Cosman
.
2003
.
Human cytomegalovirus glycoprotein UL16 causes intracellular sequestration of NKG2D ligands, protecting against natural killer cell cytotoxicity.
J. Exp. Med.
197
:
1427
1439
.
29
Chalupny
,
N. J.
,
A.
Rein-Weston
,
S.
Dosch
,
D.
Cosman
.
2006
.
Down-regulation of the NKG2D ligand MICA by the human cytomegalovirus glycoprotein UL142.
Biochem. Biophys. Res. Commun.
346
:
175
181
.
30
Lu
,
S. C.
2009
.
Regulation of glutathione synthesis.
Mol. Aspects Med.
30
:
42
59
.
31
Traverso
,
N.
,
R.
Ricciarelli
,
M.
Nitti
,
B.
Marengo
,
A. L.
Furfaro
,
M. A.
Pronzato
,
U. M.
Marinari
,
C.
Domenicotti
.
2013
.
Role of glutathione in cancer progression and chemoresistance.
Oxid. Med. Cell. Longev.
2013
: 972913.
32
Kaplowitz
,
N.
,
T. Y.
Aw
,
M.
Ookhtens
.
1985
.
The regulation of hepatic glutathione.
Annu. Rev. Pharmacol. Toxicol.
25
:
715
744
.
33
Meister
,
A.
,
M. E.
Anderson
.
1983
.
Glutathione.
Annu. Rev. Biochem.
52
:
711
760
.
34
Turrens
,
J. F.
2003
.
Mitochondrial formation of reactive oxygen species.
J. Physiol.
552
:
335
344
.
35
Sena
,
L. A.
,
S.
Li
,
A.
Jairaman
,
M.
Prakriya
,
T.
Ezponda
,
D. A.
Hildeman
,
C. R.
Wang
,
P. T.
Schumacker
,
J. D.
Licht
,
H.
Perlman
, et al
.
2013
.
Mitochondria are required for antigen-specific T cell activation through reactive oxygen species signaling.
Immunity
38
:
225
236
.
36
Ramsey
,
M. R.
,
N. E.
Sharpless
.
2006
.
ROS as a tumour suppressor?
Nat. Cell Biol.
8
:
1213
1215
.
37
Gold
,
R.
,
R. A.
Linker
,
M.
Stangel
.
2012
.
Fumaric acid and its esters: an emerging treatment for multiple sclerosis with antioxidative mechanism of action.
Clin. Immunol.
142
:
44
48
.
38
Brück
,
J.
,
R.
Dringen
,
A.
Amasuno
,
I.
Pau-Charles
,
K.
Ghoreschi
.
2018
.
A review of the mechanisms of action of dimethylfumarate in the treatment of psoriasis.
Exp. Dermatol.
27
:
611
624
.
39
Linker
,
R. A.
,
D. H.
Lee
,
S.
Ryan
,
A. M.
van Dam
,
R.
Conrad
,
P.
Bista
,
W.
Zeng
,
X.
Hronowsky
,
A.
Buko
,
S.
Chollate
, et al
.
2011
.
Fumaric acid esters exert neuroprotective effects in neuroinflammation via activation of the Nrf2 antioxidant pathway.
Brain
134
:
678
692
.
40
Gillard
,
G. O.
,
B.
Collette
,
J.
Anderson
,
J.
Chao
,
R. H.
Scannevin
,
D. J.
Huss
,
J. D.
Fontenot
.
2015
.
DMF, but not other fumarates, inhibits NF-κB activity in vitro in an Nrf2-independent manner.
J. Neuroimmunol.
283
:
74
85
.
41
Loewe
,
R.
,
W.
Holnthoner
,
M.
Groger
,
M.
Pillinger
,
F.
Gruber
,
D.
Mechtcheriakova
,
E.
Hofer
,
K.
Wolff
,
P.
Petzelbauer
. (
2002
)
Dimethylfumarate inhibits TNF-induced nuclear entry of NF-kappa B/p65 in human endothelial cells.
J. Immunol.
168
:
4781
4787
.
42
Schmidt
,
T. J.
,
M.
Ak
,
U.
Mrowietz
.
2007
.
Reactivity of dimethyl fumarate and methylhydrogen fumarate towards glutathione and N-acetyl-L-cysteine--preparation of S-substituted thiosuccinic acid esters.
Bioorg. Med. Chem.
15
:
333
342
.
43
Zheng
,
L.
,
S.
Cardaci
,
L.
Jerby
,
E. D.
MacKenzie
,
M.
Sciacovelli
,
T. I.
Johnson
,
E.
Gaude
,
A.
King
,
J. D.
Leach
,
R.
Edrada-Ebel
, et al
.
2015
.
Fumarate induces redox-dependent senescence by modifying glutathione metabolism.
Nat. Commun.
6
:
6001
.
44
Loewe
,
R.
,
T.
Valero
,
S.
Kremling
,
B.
Pratscher
,
R.
Kunstfeld
,
H.
Pehamberger
,
P.
Petzelbauer
.
2006
.
Dimethylfumarate impairs melanoma growth and metastasis.
Cancer Res.
66
:
11888
11896
.
45
Calvert
,
P.
,
K. S.
Yao
,
T. C.
Hamilton
,
P. J.
O’Dwyer
.
1998
.
Clinical studies of reversal of drug resistance based on glutathione.
Chem. Biol. Interact.
111-112
:
213
224
.
46
Reichwald
,
K.
,
T. Z.
Jørgensen
,
P.
Tougaard
,
S.
Skov
.
2014
.
TL1A induces TCR independent IL-6 and TNF-α production and growth of PLZF+ leukocytes.
PLoS One
9
: e85793.
47
Bantug
,
G. R.
,
L.
Galluzzi
,
G.
Kroemer
,
C.
Hess
.
2018
.
The spectrum of T cell metabolism in health and disease.
Nat. Rev. Immunol.
18
:
19
34
.
48
Raulet
,
D. H.
,
S.
Gasser
,
B. G.
Gowen
,
W.
Deng
,
H.
Jung
.
2013
.
Regulation of ligands for the NKG2D activating receptor.
Annu. Rev. Immunol.
31
:
413
441
.
49
Andersson
,
A. K.
,
P. F.
Sumariwalla
,
F. E.
McCann
,
P.
Amjadi
,
C.
Chang
,
K.
McNamee
,
D.
Tornehave
,
C.
Haase
,
H.
Agersø
,
V. W.
Stennicke
, et al
.
2011
.
Blockade of NKG2D ameliorates disease in mice with collagen-induced arthritis: a potential pathogenic role in chronic inflammatory arthritis.
Arthritis Rheum.
63
:
2617
2629
.
50
Huang
,
W. C.
,
N. J.
Easom
,
X. Z.
Tang
,
U. S.
Gill
,
H.
Singh
,
F.
Robertson
,
C.
Chang
,
J.
Trowsdale
,
B. R.
Davidson
,
W. M.
Rosenberg
, et al
. (
2017
)
T Cells Infiltrating Diseased Liver Express Ligands for the NKG2D Stress Surveillance System.
J. Immunol.
198
:
1172
1182
.
51
Molfetta
,
R.
,
L.
Quatrini
,
A.
Santoni
,
R.
Paolini
.
2017
.
Regulation of NKG2D-dependent NK cell functions: the yin and the yang of receptor endocytosis.
Int. J. Mol. Sci.
18
: E1677.
52
Fionda
,
C.
,
A.
Soriani
,
A.
Zingoni
,
A.
Santoni
,
M.
Cippitelli
.
2015
.
NKG2D and DNAM-1 ligands: Molecular targets for NK cell-mediated Immunotherapeutic Intervention in multiple Myeloma.
BioMed Res. Int.
2015
: 178698.
53
Yamamoto
,
K.
,
Y.
Fujiyama
,
A.
Andoh
,
T.
Bamba
,
H.
Okabe
.
2001
.
Oxidative stress increases MICA and MICB gene expression in the human colon carcinoma cell line (CaCo-2).
Biochim. Biophys. Acta
1526
:
10
12
.
54
Hagemann-Jensen
,
M.
,
F.
Uhlenbrock
,
S.
Kehlet
,
L.
Andresen
,
C.
Gabel-Jensen
,
L.
Ellgaard
,
B.
Gammelgaard
,
S.
Skov
.
2014
.
The selenium metabolite methylselenol regulates the expression of ligands that trigger immune activation through the lymphocyte receptor NKG2D.
J. Biol. Chem.
289
:
31576
31590
.
55
Cohen
,
M. B.
,
D. L.
Duvel
.
1988
.
Characterization of the inhibition of glutathione reductase and the recovery of enzyme activity in exponentially growing murine leukemia (L1210) cells treated with 1,3-bis(2-chloroethyl)-1-nitrosourea.
Biochem. Pharmacol.
37
:
3317
3320
.
56
Salati
,
L. M.
,
B.
Amir-Ahmady
.
2001
.
Dietary regulation of expression of glucose-6-phosphate dehydrogenase.
Annu. Rev. Nutr.
21
:
121
140
.
57
Yang
,
Y.
,
V.
Valera
,
C.
Sourbier
,
C. D.
Vocke
,
M.
Wei
,
L.
Pike
,
Y.
Huang
,
M. A.
Merino
,
G.
Bratslavsky
,
M.
Wu
, et al
.
2012
.
A novel fumarate hydratase-deficient HLRCC kidney cancer cell line, UOK268: a model of the Warburg effect in cancer.
Cancer Genet.
205
:
377
390
.
58
Isaacs
,
J. S.
,
Y. J.
Jung
,
D. R.
Mole
,
S.
Lee
,
C.
Torres-Cabala
,
Y. L.
Chung
,
M.
Merino
,
J.
Trepel
,
B.
Zbar
,
J.
Toro
, et al
.
2005
.
HIF overexpression correlates with biallelic loss of fumarate hydratase in renal cancer: novel role of fumarate in regulation of HIF stability.
Cancer Cell
8
:
143
153
.
59
Adam
,
J.
,
E.
Hatipoglu
,
L.
O’Flaherty
,
N.
Ternette
,
N.
Sahgal
,
H.
Lockstone
,
D.
Baban
,
E.
Nye
,
G. W.
Stamp
,
K.
Wolhuter
, et al
.
2011
.
Renal cyst formation in Fh1-deficient mice is independent of the Hif/Phd pathway: roles for fumarate in KEAP1 succination and Nrf2 signaling.
Cancer Cell
20
:
524
537
.
60
Perrier-Trudova
,
V.
,
B. W.
Huimin
,
S.
Kongpetch
,
D.
Huang
,
P.
Ong
,
A.
Le Formal
,
S. L.
Poon
,
E. Y.
Siew
,
S. S.
Myint
,
S.
Gad
, et al
.
2015
.
Fumarate hydratase-deficient cell line NCCFH1 as a new in vitro model of hereditary papillary renal cell carcinoma type 2.
Anticancer Res.
35
:
6639
6653
.
61
Cecchini
,
G.
2003
.
Function and structure of complex II of the respiratory chain.
Annu. Rev. Biochem.
72
:
77
109
.
62
Ehinger
,
J. K.
,
S.
Piel
,
R.
Ford
,
M.
Karlsson
,
F.
Sjövall
,
E. A.
Frostner
,
S.
Morota
,
R. W.
Taylor
,
D. M.
Turnbull
,
C.
Cornell
, et al
.
2016
.
Cell-permeable succinate prodrugs bypass mitochondrial complex I deficiency.
Nat. Commun.
7
:
12317
.
63
Zhao
,
T.
,
X.
Mu
,
Q.
You
.
2017
.
Succinate: an initiator in tumorigenesis and progression.
Oncotarget
8
:
53819
53828
.
64
Pollard
,
P. J.
,
M.
El-Bahrawy
,
R.
Poulsom
,
G.
Elia
,
P.
Killick
,
G.
Kelly
,
T.
Hunt
,
R.
Jeffery
,
P.
Seedhar
,
J.
Barwell
, et al
.
2006
.
Expression of HIF-1alpha, HIF-2alpha (EPAS1), and their target genes in paraganglioma and pheochromocytoma with VHL and SDH mutations.
J. Clin. Endocrinol. Metab.
91
:
4593
4598
.
65
Alvarez
,
M.
,
F.
Simonetta
,
J.
Baker
,
A.
Pierini
,
A. S.
Wenokur
,
A. R.
Morrison
,
W. J.
Murphy
,
R. S.
Negrin
.
2019
.
Regulation of murine NK cell exhaustion through the activation of the DNA damage repair pathway.
JCI Insight
5
: 127729.
66
Warburg
,
O.
1956
.
On the origin of cancer cells.
Science
123
:
309
314
.
67
Yan
,
H.
,
D. W.
Parsons
,
G.
Jin
,
R.
McLendon
,
B. A.
Rasheed
,
W.
Yuan
,
I.
Kos
,
I.
Batinic-Haberle
,
S.
Jones
,
G. J.
Riggins
, et al
.
2009
.
IDH1 and IDH2 mutations in gliomas.
N. Engl. J. Med.
360
:
765
773
.
68
O'Brien
,
M. L.
,
Tew
,
K. D.
(
1996
)
Glutathione and related enzymes in multidrug resistance.
Eur.J. Cancer
32A
:
967
978
.
69
Zhang
,
X.
,
A.
Rao
,
P.
Sette
,
C.
Deibert
,
A.
Pomerantz
,
W. J.
Kim
,
G.
Kohanbash
,
Y.
Chang
,
Y.
Park
,
J.
Engh
, et al
.
2016
.
IDH mutant gliomas escape natural killer cell immune surveillance by downregulation of NKG2D ligand expression.
Neuro-oncol.
18
:
1402
1412
.
70
Godwin
,
A. K.
,
A.
Meister
,
P. J.
O’Dwyer
,
C. S.
Huang
,
T. C.
Hamilton
,
M. E.
Anderson
.
1992
.
High resistance to cisplatin in human ovarian cancer cell lines is associated with marked increase of glutathione synthesis.
Proc. Natl. Acad. Sci. USA
89
:
3070
3074
.
71
Hochwald
,
S. N.
,
D. M.
Rose
,
M. F.
Brennan
,
M. E.
Burt
.
1997
.
Elevation of glutathione and related enzyme activities in high-grade and metastatic extremity soft tissue sarcoma.
Ann. Surg. Oncol.
4
:
303
309
.
72
Quinlan
,
C. L.
,
A. L.
Orr
,
I. V.
Perevoshchikova
,
J. R.
Treberg
,
B. A.
Ackrell
,
M. D.
Brand
.
2012
.
Mitochondrial complex II can generate reactive oxygen species at high rates in both the forward and reverse reactions.
J. Biol. Chem.
287
:
27255
27264
.
73
He
,
W.
,
F. J.
Miao
,
D. C.
Lin
,
R. T.
Schwandner
,
Z.
Wang
,
J.
Gao
,
J. L.
Chen
,
H.
Tian
,
L.
Ling
.
2004
.
Citric acid cycle intermediates as ligands for orphan G-protein-coupled receptors.
Nature
429
:
188
193
.
74
Rubic
,
T.
,
G.
Lametschwandtner
,
S.
Jost
,
S.
Hinteregger
,
J.
Kund
,
N.
Carballido-Perrig
,
C.
Schwärzler
,
T.
Junt
,
H.
Voshol
,
J. G.
Meingassner
, et al
.
2008
.
Triggering the succinate receptor GPR91 on dendritic cells enhances immunity.
Nat. Immunol.
9
:
1261
1269
.
75
Littlewood-Evans
,
A.
,
S.
Sarret
,
V.
Apfel
,
P.
Loesle
,
J.
Dawson
,
J.
Zhang
,
A.
Muller
,
B.
Tigani
,
R.
Kneuer
,
S.
Patel
, et al
.
2016
.
GPR91 senses extracellular succinate released from inflammatory macrophages and exacerbates rheumatoid arthritis.
J. Exp. Med.
213
:
1655
1662
.

The authors have no financial conflicts of interest.

Supplementary data