Branched-chain amino acid transaminase 1 (BCAT1) is upregulated selectively in human isocitrate dehydrogenase (IDH) wildtype (WT) but not mutant glioblastoma multiforme (GBM) and promotes IDHWT GBM growth. Through a metabolic synthetic lethal screen, we report here that α-ketoglutarate (AKG) kills IDHWT GBM cells when BCAT1 protein is lost, which is reversed by reexpression of BCAT1 or supplementation with branched-chain α-ketoacids (BCKA), downstream metabolic products of BCAT1. In patient-derived IDHWT GBM tumors in vitro and in vivo, cotreatment of BCAT1 inhibitor gabapentin and AKG resulted in synthetic lethality. However, AKG failed to evoke a synthetic lethal effect with loss of BCAT2, BCKDHA, or GPT2 in IDHWT GBM cells. Mechanistically, loss of BCAT1 increased the NAD+/NADH ratio but impaired oxidative phosphorylation, mTORC1 activity, and nucleotide biosynthesis. These metabolic alterations were synergistically augmented by AKG treatment, thereby causing mitochondrial dysfunction and depletion of cellular building blocks, including ATP, nucleotides, and proteins. Partial restoration of ATP, nucleotides, proteins, and mTORC1 activity by BCKA supplementation prevented IDHWT GBM cell death conferred by the combination of BCAT1 loss and AKG. These findings define a targetable metabolic vulnerability in the most common subset of GBM that is currently incurable.

Significance:

Metabolic synthetic lethal screening in IDHWT glioblastoma defines a vulnerability to ΑΚG following BCAT1 loss, uncovering a therapeutic strategy to improve glioblastoma treatment.

See related commentary by Meurs and Nagrath, p. 2354

Glioblastoma multiforme (GBM), which represents about 15% of all brain tumors, is a highly malignant grade IV brain tumor that dominantly expresses wildtype isocitrate dehydrogenase (IDHWT; ref. 1). With current standard of treatment consisting of maximal surgical resection followed by radiotherapy and temozolomide chemotherapy, median survival of patients with IDHWT GBM is only about 15 months (1), which highlights an unmet need for new therapies that can better treat this aggressive disease. Synthetic lethality was initially described in Drosophila melanogaster in 1922 and has received much enthusiasm for the treatment of cancer by selectively killing cancer cells with simultaneous loss of two interacting survival genes or pathways (2). Although a couple of synthetic lethal interactions with altered metabolism in mutant IDH (IDHMut) gliomas have been reported previously (3, 4), it remains unknown whether metabolic synthetic lethality can be applied to treat IDHWT GBM.

Amino acids are one of the most fundamental bricks of cell structure and essential for synthesis of protein and nucleotide to support cell proliferation. Recent studies have revealed the significance of altered branched-chain amino acid (BCAA) metabolism in various human cancers including GBM, pancreatic cancer, lung cancer, breast cancer, leukemia, kidney cancer, ovarian cancer, and liver cancer (5). BCAAs (i.e., leucine, isoleucine, and valine) belong to a group of essential amino acids and are reversibly catabolized by cytosolic branched-chain aminotransferase (BCAT) 1 and mitochondrial BCAT2 into branched-chain α-ketoacids (BCKA), including α-ketoisocaproate (KIC), α-ketoisovalerate (KIV), α-keto-β-methylvalerate (KMV; ref. 5). Meanwhile, their amino groups are transferred to α-ketoglutarate (AKG) to produce glutamate, and thus BCAAs are the critical nitrogen donors for synthesis of nucleotides and nonessential amino acids in cancer cells (6). Studies from early pancreatic cancer, lung cancer, leukemia, and liver cancer showed that reductive metabolism of BCAAs is dominant in these cancer cells, leading to accumulation of BCAAs and subsequent mTORC1 activation (6–11). mTORC1 is a master regulator of protein synthesis and metabolic reprogramming and, upon activation, promotes cancer cell proliferation (12). In contrast, the BCAA catabolic pathway is highly active in leukemia stem cells and IDHWT GBM (13–15). High expression of BCAT1 increases glutamate production but reduces intracellular levels of AKG in leukemia cells, which causes activation of hypoxia-inducible factor 1α (HIF1α) and DNA hypermethylation (15), two important oncogenic drivers of tumor growth. Interestingly, none or very little (1%–3% in pancreatic cancer) of the final metabolic products in the BCAA metabolic pathway contributes to carbons for the tricarboxylic acid cycle in cancer cells (6–8), strongly arguing that the intermediates produced from BCAA metabolism may play a critical role in cancer cell proliferation. BCAT1 is upregulated in IDHWT GBM, which is controlled by its promoter hypomethylation and hypoxia (13, 14), whereas both BCAT1 and BCAT2 are inactivated by R-2-hydroxyglutarate (HG) in IDHMut glioma (4). BCAT1 promotes growth of IDHWT GBM in vitro and in vivo (13, 14), suggesting that BCAT1 is an intriguing target for the development of metabolic synthetic lethality in IDHWT GBM.

In the current study, we reported that genetic or pharmacologic inhibition of BCAT1 (BCAT1i) in combination of AKG kills IDHWT GBM cells in vitro and in vivo. Supplementation with KIC prevents BCAT1i and AKG-induced mitochondrial dysfunction and depletion of ATP, nucleotides, and proteins, leading to IDHWT GBM cell survival. These findings define a novel metabolic vulnerability in the most common and currently incurable subset of GBM and develop an innovative and nontoxic therapeutic strategy.

Plasmid constructs

Full-length human BCAT1 cDNA was amplified by PCR and cloned into pcFUGW-3XFLAG vector. DNA oligonucleotides of the single-guide RNA targeting human BCAT1, BCAT2, BCKA dehydrogenase E1 subunit α (BCKDHA), or GPT2 (Supplementary Table S1) were annealed and ligated into BsmBI-linearized lentiCRISPRv2 vector (Addgene, #52961). All recombinant plasmids were verified by Sanger sequencing.

Cell culture

U251MG (RRID: CVCL_0021), LN229 (RRID: CVCL_0393; gifts from Sandeep Burma at UT Health San Antonio in 2015), U87MG (RRID: CVCL_0022, gift from Gregg L. Semenza at Johns Hopkins in 2014), T98G (RRID: CVCL_0556), and LN18 (RRID: CVCL_0392; purchased from ATCC in 2017) cells were cultured in DMEM supplemented with 5% or 10% heat-inactivated FBS at 37°C in a 5% CO2/95% air incubator. Patient-derived IDHWT GBM C116 cells were generated by J.A. Copland. and cultured in DMEM/Ham's F-12 supplemented with 1% nonessential amino acid, 1 nmol/L T3, 8 ng/mL EGF, 5 μg/mL ITS, 5% iron-supplemented bovine calf serum, and 1% penicillin-streptomycin at 37°C in a 5% CO2/95% air incubator. Primary IDHWT GBM tumorspheres were cultured as described previously (13), which was approved by the Institutional Review Board at UT Southwestern Medical Center (Dallas, TX) with written informed consent. Primary mouse astrocytes were isolated from the cortex of 1- to 4-day old C57BL/6J mice and cultured in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin at 37°C in a 5% CO2/95% air incubator. Microglia and oligodendrocyte precursor cells were removed by sequential shaking at 180 rpm for 30 minutes and 240 rpm for 6 hours. Astrocytes at the second passage were used for the experiment. Hypoxia was conducted by exposure of cells to 1% O2/5% CO2/balanced N2 gas in a modular incubator chamber (Billups-Rothenberg). All knockout (KO) cell lines were generated using the CRISPR/Cas9 technique as described previously (13). All cell lines used for experiments were replaced by thawing frozen cells every month. All cell lines and C116 cells were annually tested to be Mycoplasma free and authenticated by short tandem repeat DNA profiling analysis during 2015–2017 (Supplementary Table S2).

Clonogenic assay

Cells were seeded on a 48-well plate and treated with indicated chemicals for 7 or 12 days. Colonies were washed with PBS, fixed with 100% methanol, and stained with 0.01% crystal violet. The crystal violet dye was dissolved in 10% acetic acid and measured at optical density (OD) 570 nm in a Tecan Spark 10M plate reader.

Cell viability assay

Cells (4,000 cells/well) were seeded on a 96-well plate and treated with indicated chemicals for 2 days. Cell viability was measured by the CellTiter 96 AQueous Non-Radioactive Cell Proliferation Assay kit (catalog no.: G5421, Promega). For propidium iodide staining, cells were stained with 2 μmol/L propidium iodide for 10 minutes and imaged under a Zeiss Axio Observer Z1 microscope.

Metabolomics assay

Parental and BCAT1 KO#2 U251MG cells were seeded on 6-cm dishes and treated with vehicle or cell-permeable AKG (10 mmol/L) for 24 hours. After washing twice with cold PBS, cells were lysed by 1 mL of cold extraction buffer (80% methanol/20% water/0.2 μmol/L internal heavy standard mix), vortexed for 30 seconds, and centrifuged at 13,000 rpm for 15 minutes at 4°C. The supernatant containing intracellular metabolites was filtered using a polyvinylidene difluoride syringe filter (catalog no.: F2504-6, Thermo Fisher Scientific) and analyzed by LC/MS. The pellet was dissolved in 0.1 N NaOH and their protein concentration was quantified by Bradford assay for metabolite normalization. The heatmap was generated by the statistical analysis module on the MetaboAnalyst website.

For nucleotide quantification, parental and BCAT1 KO#3 U251MG cells were treated with vehicle, cell-permeable AKG (10 mmol/L), KIC (1 mmol/L), or both for 24 hours. After washing with ice-cold saline, cells were lysed by 1 mL of 80% acetonitrile mixed with 1 μmol/L Adenosine-15N5 5′-monophosphate disodium salt. Intracellular metabolites were extracted by repeated freeze-thaw cycles, followed by centrifugation at 20,160 × g for 15 minutes. The supernatant containing nucleotides was filtered through a pretreated SPE cartridge and analyzed by LC/MS. The pellet was dissolved in 0.1 N NaOH and their protein concentration was quantified by Bradford assay for nucleotide normalization.

Mitochondrial complex activity assay

Parental and BCAT1 KO cells were harvested, washed once with ice-cold PBS, and resuspended in isolation buffer (250 mmol/L sucrose, 10 mmol/L KCl, 1.5 mmol/L MgCl2, 1 mmol/L Na2EDTA, 1 mmol/L dithiothreitol, 1 mmol/L phenylmethylsulfonylfluoride (PMSF), 10 mmol/L Tris-HCl, pH 7.4, and protease inhibitor mixture). After incubation on ice for 30 minutes, cells were homogenized and subjected to centrifugation at 100 × g for 5 minutes at 4°C to remove cells that are not fractionated. The supernatant was collected and centrifuged twice at 700 × g for 10 minutes at 4°C to remove the nuclear fraction. The resulting supernatant was subjected to centrifugation at 12,000 × g for 15 minutes at 4°C and the pellet as the mitochondrial fraction was collected and washed once with isolation buffer. Mitochondria were resuspended in lysis buffer (30 mmol/L Tris-HCl pH7.4, 200 mmol/L KCl, 5 mmol/L EDTA, 0.5% Triton X-100, 1 mmol/L PMSF, and protease inhibitor mixture), and subjected to freeze in liquid nitrogen and thaw for three rounds to prepare mitochondrial extracts. The activity of mitochondrial complex I was determined through oxidation of NADH with ubiquinone as the electron acceptor. Mitochondrial extracts (20 μg) were mixed with the assay buffer (25 mmol/L KHPO4, 25 mmol/L K2HPO4, pH 7.2, 2.5 mg/mL BSA, 5 mmol/L MgCl2, 2 mmol/L KCN, 0.01 mg/mL antimycin A, 1.3 mmol/L NADH, 0.01 mmol/L decylubiquinone) prewarmed for 5 minutes at 30°C. The absorbance at 340 nm wavelength was monitored every 30 seconds for 10 minutes in a plate reader at 37°C after the stable baseline was achieved. 0.25 mmol/L rotenone (10 μL) was added to stop the reaction and the absorbance at 340 nm was monitored for additional 5 minutes. The activity of mitochondrial complex V was determined through the NADH oxidation via conversion of phosphoenolpyruvate to lactate by two-step reaction. Mitochondrial extracts (20 μg) were mixed with 2.5 mmol/L ATP in the assay buffer (40 mmol/L Tris-HCO3, 2.5 mol/L phosphoenolpyruvate, 0.2 mmol/L NADH, 0.025 mg/mL antimycin A, 50 mmol/L MgCl2, 0.5 mg/mL lactate dehydrogenase, 0.1 mg/mL pyruvate kinase) prewarmed for 5 minutes at 30°C. The absorbance at 340 nm wavelength was monitored every 30 seconds for 10 minutes in a plate reader at 37°C after the stable baseline was achieved. Both activities of mitochondrial complex I and V were normalized to citrate synthase activity (16).

Tetramethylrhodamine, ethyl ester assay

Cells (2 × 105 cells/well) were seeded into a 12-well plate. Next day, cells were stained with 50 nmol/L tetramethylrhodamine, ethyl ester (TMRE) for 30 minutes at 37°C in 5% CO2 incubator. Fluorescence images were taken with a motorized Axio Observer Z1 microscope (Carl Zeiss). For negative control, cells were pretreated with 10 μmol/L carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) 30 minutes before TMRE staining. Intensity of fluorescence was analyzed by Image J.

Reactive oxygen species assay

Cells (5 × 105 cells/well) were seeded into a 6-well plate. Next day, cells were treated with MitoSOX (2 μmol/L) for 1 hour at 37°C in 5% CO2 incubator, dissociated with trypsin, washed once, and resuspended in 0.5 mL PBS. Flow cytometry was performed to quantify reactive oxygen species (ROS) levels.

NAD+/NADH assay

The cellular NAD+ and NADH levels were measured using NAD/NADH-Glo assay (catalog no.: G9071, Promega). Cells (1 × 105 cells/well) were seeded into a 24-well plate. After 48 hours, cells were lysed by adding 300 μL/well of 0.1 mol/L NaOH with 0.5% DTAB and subjected to measurement in a Tecan Spark 10M plate reader according to manufacturer's instructions. The NAD+ and NADH levels were normalized to cellular protein concentrations quantified by Bradford assay (Bio-Rad).

ATP assay

The cellular ATP levels were measured using an ATP determination kit (catalog no.: A22066, Thermo Fisher Scientific). Cells (1 × 105 cells/well) were seeded into a 24-well plate. After 24 hours, cells were lysed by adding of 500 μL of 1% Triton X-100 and subjected to measurement in a Tecan Spark 10M plate reader according to manufacturer's instructions. The ATP levels were normalized to cellular protein concentrations quantified by Bradford assay (Bio-Rad).

Seahorse assay

The oxygen consumption rate (OCR) was measured in a Seahorse XF24e extracellular flux analyzer. Cells were seeded at 3–9 × 104/well into a V7 seahorse 24-well plate and next day subjected to Seahorse analysis. Oligomycin A (1 μmol/L), FCCP (350 nmol/L), and rotenone (2 μmol/L) were added into different ports of the Seahorse cartridge. OCR was measured with a standard 8-minute cycling program including mixture (3 minutes), waiting (2 minutes), and measurement (3 minutes). After Seahorse analysis, cells were lysed in 0.1 mol/L NaOH and cellular protein concentrations were determined by Bradford assay (Bio-Rad) for data normalization.

AKG assay

The cellular AKG levels were measured using a AKG colorimetric/fluorometric assay kit (catalog no.: K677-100, Biovision). A total of 2 × 106 of cells were rapidly lysed with 150 μL of ice-cold AKG assay buffer and then sonicated. After centrifugation at 13,000 rpm for 10 minutes, supernatants were filtered using a 10 kDa spin column (Millipore) and subjected to measurement at 535/587 nm in a Tecan Spark 10M plate reader according to manufacturer's instructions. The AKG levels were normalized to cellular protein concentrations quantified by Bradford assay (Bio-Rad).

Immunostaining assay

Cells were seeded at 5 × 104 cells per well onto glass coverslips placed in a 12-well plate. Next day, cells were treated with cell-permeable AKG (10 mmol/L) or DMSO for 24 hours. After washing once with PBS, cells were fixed with 4% paraformaldehyde for 20 minutes at room temperature, permeabilized with 0.1% Triton X-100 in PBS for 15 minutes, and blocked with 5% BSA in PBS for 60 minutes. Cells then were incubated overnight with antimitochondria antibody (MTC02); 1:50 dilution in PBS with 1% BSA (catalog no.: ab3298, Abcam, RRID: AB_303683) or anti-FLAG antibody (1:200 dilution in PBS with 1% BSA, RRID: AB_259529) in a 12-well plate at 4°C, washed with PBST (PBS with 0.1% Tween-20) for three times, incubated for 90 minutes with Alexa488 donkey anti-mouse IgG (1:200 dilution in PBST with 1% BSA) or Cy3 donkey anti-mouse IgG (1:1,000 dilution in PBST with 1% BSA) in dark, washed again with PBST for three times, and incubated for 5 minutes with DAPI (1:1,000 dilution in PBS) in dark. After washing three times, cells were mounted with antifade mounting medium. Mounted slides were observed with a Zeiss Axio Observer Z1 fluorescence microscope.

Transmission electron microscope assay

Cells were fixed with 2.5% (volume for volume) glutaraldehyde in 0.1 mol/L sodium cacodylate buffer (pH 7.4) for 2 hours at room temperature. After washing with 0.1 mol/L sodium cacodylate buffer (pH 7.4) five times, cells were postfixed in 1% osmium tetroxide with 0.8% K3Fe(CN6) for 1.5 hours at room temperature. After rinsing with water five times, cells were stained with 4% uranyl acetate for 2 hours at room temperature in dark, dehydrated with increasing concentrations of ethanol ranging from 50% to 95%, infiltrated with Embed-812 resin, and polymerized overnight in a 70°C oven. Samples were sectioned with a Leica UCT ultramicrotome, placed onto copper grids, and stained with 2% uranyl acetate. Images were acquired using a JEOL 1400 Plus transmission electron microscope equipped with an AMT CCD camera.

Immunoblot assay

Cells were lysed in modified lysis buffer (50 mmol/L Tris-HCl, pH 7.5, 150 mmol/L NaCl, 1 mmol/L β-mercaptoethanol, 1% Igepal, and protease inhibitor cocktail) as described previously (13). Equal amounts of lysates were fractionated by SDS-PAGE and subjected to immunoblot assay with the following antibodies: BCAT1 (Proteintech, catalog no.: 13640-1-AP, RRID: AB_2063569), BCAT2 (ProteinTech, catalog no.: 16417-1-AP, RRID: AB_10792411), GPT2 (Proteintech, catalog no.: 16757-1-AP, RRID: AB_2112098), BCKDHA (Bethyl Laboratories, catalog no. A303-790A, RRID: AB_11218185), p-AMPK (Cell Signaling Technology, catalog no.: 2535, RRID: AB_331250), AMPK (Cell Signaling Technology, catalog no.: 2532S, RRID: AB_330331), p-P70S6K (Cell Signaling Technology, catalog no.: 9205S, RRID: AB_330944), P70S6K (Cell Signaling Technology, catalog no.: 9202S, RRID: AB_331676), or actin (ProteinTech, catalog no.: 66009-1-Ig, RRID: AB_2687938).

Protein synthesis assay

Cells were seeded at 50% confluency in a 6-well plate, and treated with indicated chemicals for 24–48 hours. Before harvesting for lysis, cells were treated with 1 μmol/L puromycin dihydrochloride (catalog no.: P8833, Sigma) for 30 minutes. Puromycin incorporation into proteins was detected by immunoblot assay with anti-puromycin antibody (1:5,000 dilution, catalog no.: MABE343, Sigma, RRID: AB_2566826).

Animal studies

Animal experiments were approved by the Animal Care and Use Committee at UT Southwestern Medical Center (Dallas, TX). IDHWT GBM patient-derived xenograft (PDX) tumors were obtained from the brain tumor PDX national resource at Mayo Clinic and approved by the Institutional Review Board at UT Southwestern Medical Center. PDX tumors were cut into 8 mm3 and implanted subcutaneously into the right flank of male NOD-SCIDIL2rγnull (NSG) mice. A total of 100 μL Matrigel was injected into the area of implantation right after. Once the tumor volume reached 50–100 mm3, mice were randomly grouped and administered with saline, 150 mg/kg AKG disodium salt (prepared in saline solution), 60 mg/kg gabapentin (prepared in saline solution), or their combination by oral gavage daily for 16 days. Tumor growth was monitored with a caliper every 3 days and calculated with the formula: volume = 0.52 × length × height × width.

IHC assay

IHC assays were performed by the Dako Autostainer Link 48 system as described previously (17). Briefly, the slides were baked at 60°C for 20 minutes, deparaffinized and hydrated. After the antigen retrieval at optimized pH for 20 minutes in a Dako PT Link, the tissues were incubated with anti-Ki67 antibody (ProteinTech, 1:1,000, RRID: AB_2756525) for 20 minutes, and visualized using the EnVision FLEX visualization system (Dako). The percentage of Ki67-positive cells per field was counted manually.

Statistical analysis

Statistical analysis was performed by two-tailed Student t test between two groups, and one-way or two-way ANOVA with multiple testing correction within multiple groups. P < 0.05 is considered significant. Data were expressed as mean ± SEM.

Data availability

The data generated in this study are available within the article and its Supplementary Data file.

A metabolic screen identifies AKG and BCAT1i as a synthetic lethal approach in IDHWT GBM cells

To perform a BCAT1-dependent synthetic lethal screening, we first generated two independent BCAT1 KO U251MG cell lines using the CRISPR/Cas9 technique and their rescued cells by lentiviral transduction (Supplementary Fig. S1A and S1B). The levels of intracellular AKG, an amino group acceptor coupled with BCAT1-mediated transamination, were significantly elevated in BCAT1 KO#2 U251MG cells (Supplementary Fig. S1C), whereas glutamate, the product of AKG, was significantly decreased in these KO cells (Supplementary Fig. S1D). Reexpression of FLAG-BCAT1 in BCAT1 KO#2 U251MG cells reversed AKG levels comparable with those in parental U251MG cells (Supplementary Fig. S1A–S1C). These data indicate that BCAT1 KO U251MG cells we generated display the on-target loss-of-function effect on BCAA metabolism.

To identify metabolic synthetic lethal partners of BCAT1i in IDHWT GBM cells, we screened BCAA metabolism-related metabolites in BCAT1 KO#2 U251MG cells. The clonogenic assay demonstrated that dimethyl AKG (DMAKG, 10 mmol/L), a cell-permeable precursor of AKG, almost completely inhibited growth of BCAT1 KO#2 cells but had a modest effect on parental U251MG cells (Fig. 1A). In contrast, BCAAs (2 mmol/L), dimethyl (DM)-glutamate (4 mmol/L), DM-succinate (10 mmol/L), or L-2-HG (5 mmol/L) did not have a lethal effect on BCAT1 KO#2 cells (Fig. 1A). Given the critical role of redox homeostasis and mitochondrial respiration in IDHWT GBM (18, 19), we also included antioxidants and respiratory inhibitors N-acetyl cysteine (1 mmol/L), L-ascorbic acid (500 μmol/L), MitoQ (250 nmol/L), glutathione (3 mmol/L), nicotinamide mononucleotide (NMN, 1 mmol/L), rotenone (0.1 μmol/L), and oligomycin (1 μmol/L) in our screening. These compounds did not have a synergistic effect with BCAT1 KO on growth inhibition of U251MG cells (Fig. 1A). The glycolysis-related inhibitor 2-deoxy-D-glucose (0.5 mmol/L) also failed to enhance growth inhibition of BCAT1 KO#2 U251MG cells (Fig. 1A). Likewise, the ketones, 3-hydroxybutyrate (10 mmol/L) and acetoacetate (2 mmol/L), had no effect on growth of parental and BCAT1 KO#2 U251MG cells (Fig. 1A). Together, the metabolic screening reveals that cell-permeable AKG induces a lethal vulnerability in BCAT1 KO IDHWT GBM cells and that the lethal phenotype is specifically caused by AKG but not metabolites from its oxidation and transamination.

To validate the screening data, we performed clonogenic assay in another BCAT1 KO#3 U251MG cell line and observed an identical synergistic growth inhibition of BCAT1 KO cells in the presence of cell-permeable AKG (Fig. 1B). A similar lethal phenotype was also observed in BCAT1 KO#1, KO#2, and KO#3 U87MG cells (Fig. 1C). Cell death assay using propidium iodide staining confirmed that cell-permeable AKG selectively killed BCAT1 KO#2 or KO#3 but not parental U251MG cells (Fig. 1D). To determine whether cell-permeable AKG specifically kills BCAT1 KO cells, we first performed the rescued experiment in U251MG cells. Ectopic expression of BCAT1 significantly restored survival and growth of BCAT1 KO#2 and KO#3 cells under conditions of cell-permeable AKG treatment (Fig. 1,B and D; Supplementary Fig. S2A). We next found that cell-permeable AKG had no inhibitory effect on BCAT2 KO U251MG cells (Supplementary Fig. S3A and S3B). Cell-permeable AKG also failed to exert a combined lethal effect with KO of BCKDHA or another AKG-dependent transaminase GPT2 in U251MG cells (Supplementary Fig. S3C–S3G). Collectively, these results reveal the lethal effect of AKG specifically in BCAT1 KO IDHWT GBM cells in vitro.

The dose–response studies with increasing doses of cell-permeable AKG showed that growth of BCAT1 KO#2 and KO#3 U251MG cells was gradually decreased (Fig. 1,E and F). We next studied whether cell-permeable AKG can synergize with a pharmacologic BCAT1 inhibitor gabapentin to induce a lethal effect on IDHWT GBM cells. Cotreatment of cell-permeable AKG (10 mmol/L) and gabapentin (5 or 20 mmol/L) caused significantly more U251MG cell death than their individual treatment (Fig. 1,G and H; Supplementary Fig. S2B). Moreover, U251MG cell death by this combined treatment was dose dependent (Fig. 1I). Similar synergistic cell death was also observed in many other IDHWT GBM U87MG, LN229, and T98G cells highly expressing BCAT1 (Fig. 1,JM). Importantly, cell-permeable AKG alone but not gabapentin was sufficient to induce death of BCAT1-null LN18 and patient-derived IDHWT GBM C116 cells (Fig. 1,N and O), indicating the specific inhibitory effect of gabapentin on BCAT1 in IDHWT GBM cells. Finally, we found that the killing effect of gabapentin/AKG combination was not applicable to normal mouse astrocytes, and the combination instead increased astrocyte survival (Supplementary Fig. S3H). Together, these findings from genetic and pharmacologic studies indicate that BCAT1i combined with AKG induces a dose-dependent killing of IDHWT GBM cells but not normal astrocytes in vitro.

BCAT1i/AKG combination induces synthetic lethality in human patient-derived IDHWT GBM in vitro and in mice

To study potential translational benefits of BCAT1i/AKG in patients with IDHWT GBM, we treated patient-derived IDHWT GBM tumorspheres with gabapentin (5 mmol/L), cell-permeable AKG (10 mmol/L), or their combination for 7 days. Treatment of gabapentin or cell-permeable AKG alone had minimal effects on growth of tumorspheres in all four human IDHWT GBM models (Fig. 2A). In contrast, gabapentin and AKG combination robustly inhibited growth of tumorspheres in vitro (Fig. 2A), indicating that BCAT1i/AKG combination shows the killing effect in human IDHWT GBM in vitro. We next studied whether gabapentin and AKG combination induces killing of IDHWT GBM in mice. To this end, human patient-derived IDHWT GBM tumors were subcutaneously implanted into NSG mice and mice were administrated daily with gabapentin (60 mg/kg), AKG (150 mg/kg), or both by oral gavage once the tumor volume reached to 50–100 mm3. The combined treatment of gabapentin and AKG significantly inhibited growth of IDHWT GBM, whereas their individual treatment had no effect on tumor growth in mice (Fig. 2B). This combination treatment did not show any toxicity as mouse body weight was not affected (Fig. 2C). Analysis of Ki67 IHC staining revealed that gabapentin and AKG combination significantly reduced proliferating GBM tumor cells in PDX tumors as compared with saline, whereas individual treatment of gabapentin or AKG failed to do so (Fig. 2D). Together, these in vivo results from patient-derived IDHWT GBM models indicate that BCAT1i/AKG combination reduces the rate of IDHWT GBM growth in mice.

Multiple cell death inhibitors fail to prevent BCAT1i/AKG-induced IDHWT GBM cell death

Next, we studied which cell death pathway is responsible for BCAT1i/AKG-induced IDHWT GBM cell death. Various cell death inhibitors including Z-Vad (10 μmol/L, apoptosis inhibitor), ferrostatin-1 (2 μmol/L, ferroptosis inhibitor), necrostatin-1 (20 μmol/L, necroptosis inhibitor), 3-methyladenine (1 mmol/L, autophagy inhibitor), or DPQ (30 μmol/L, parthanatos inhibitor) were applied with or without cell-permeable AKG into parental and BCAT1 KO#2 and KO#3 U251MG cells. Neither of these inhibitors rescued survival of BCAT1 KO U251MG cells in the presence of cell-permeable AKG (Supplementary Fig. S4A–S4E). These findings indicate that the known cell death inhibitors cannot prevent BCAT1i/AKG-induced killing of IDHWT GBM cells.

Supplementation with BCKA prevents BCAT1i/AKG-induced IDHWT GBM cell death

Next, we studied whether BCAT1-dependent metabolites can eliminate the synergy from BCAT1i/AKG combination to prevent IDHWT GBM cell death. BCAT1 reversibly converts BCAAs into BCKAs in the cell (Fig. 3A). We first studied by clonogenic assay whether supplementation with BCAAs can prevent BCAT1i/AKG-induced cell death. BCAT1 KO cells were used for the following mechanistic studies because of their specificity (Supplementary Fig. S1). BCAAs including Leu, Ile, and Val (2 mmol/L) failed to restore survival of BCAT1 KO#2 and KO#3 U251MG cells under conditions of cell-permeable AKG treatment (Fig. 3B). Similarly, cell-permeable glutamate and glutamine did not show the rescued effect (Supplementary Fig. S5A). In contrast, supplementation with KIC (1 mmol/L) significantly reversed death of BCAT1 KO U251MG or U87MG cells conferred by cell-permeable AKG (Fig. 3C; Supplementary Fig. S5B). KIV (1 mmol/L) and KMV (1 mmol/L) also prevented AKG-induced death of BCAT1 KO#2 and KO#3 U251MG cells to a certain degree (Fig. 3C). The individual BCKA itself had no obvious effect on parental and BCAT1 KO cell growth in the absence of cell-permeable AKG (Fig. 3C; Supplementary Fig. S5B). Mass spectrometry analysis showed that all BCKAs were reduced in BCAT1 KO#2 and KO#3 U251MG cells (Fig. 3D). Together, these results indicate that the depletion of BCKAs in BCAT1-null IDHWT GBM cells promotes a synergy with AKG treatment to induce cell death.

A previous study reported that dimethyl AKG induces HIF1α protein expression in cancer cells (20). Consistently, both HIF1α and HIF2α protein levels were elevated in parental and BCAT1 KO#2 or KO#3 U251MG cells treated with cell-permeable AKG for 24 hours (Supplementary Fig. S6A). To determine the role of HIF in BCAT1i/AKG-induced IDHWT GBM cell death, we performed clonogenic assays under 20% or 1% O2 and found that hypoxia had no obvious effect on viability of parental and BCAT1 KO#2 or KO#3 U251MG cells with or without cell-permeable AKG treatment (Supplementary Fig. S6B and S6C), which excludes a possible role of HIF signaling in BCAT1i/AKG-induced IDHWT GBM cell death.

A synergistic increase in NAD+/NADH by BCAT1i/AKG combination causes IDHWT GBM cell death

BCKA oxidation by BCKA acid dehydrogenase is coupled with NAD+ reduction into NADH in mitochondria, which controls electron transfer in the mitochondrial complex I to regulate mitochondrial respiration. We next studied whether BCAT1i increases the NAD+/NADH ratio and whether this altered ratio is augmented by AKG treatment to block mitochondrial respiration, leading to IDHWT GBM cell death. Our biochemical studies showed that cellular NAD+ levels were significantly increased but cellular NADH levels were decreased in BCAT1 KO#2 and KO#3 U251MG cells (Supplementary Fig. S7A and S7B). As a result, the ratio of NAD+/NADH was significantly increased in BCAT1 KO U251MG cells (Supplementary Fig. S7C). Consistently, BCAT1 KO#2 and KO#3 significantly reduced the activity of the mitochondrial complex I in U251MG cells (Supplementary Fig. S7D). Imbalance of NAD+/NADH ratio is associated with redox homeostasis in mitochondria (21). As shown by MitoSOX-based flow cytometry, the mitochondrial ROS levels were significantly increased in BCAT1 KO#2 and KO#3 U251MG cells as compared with parental cells (Supplementary Fig. S7E). Seahorse analysis demonstrated that BCAT1 KO#3 inhibited the basal OCR as well as maximal respiratory capacity in mitochondria (Supplementary Fig. S7F). The OCR was also inhibited when BCKDHA was knocked out in U251MG cells (Supplementary Fig. S7G), which supports the idea that reduced BCKA oxidation increases the NAD+/NADH ratio to inhibit oxidative phosphorylation in BCAT1 KO cells. Consistently, the cellular ATP levels were significantly decreased in BCAT1 KO#2 and KO#3 U251MG cells (Supplementary Fig. S7H). TMRE staining revealed that both BCAT1 KO#2 and KO#3 significantly reduced mitochondrial membrane potential in U251MG cells (Supplementary Fig. S7I and S7J), which further support the role of BCAT1 in mitochondrial ATP production. However, the activity of the mitochondrial complex V ATP synthase was not altered in BCAT1 KO#2 and KO#3 U251MG cells (Supplementary Fig. S7K). Collectively, these results indicate that BCAT1 loss inhibits NADH/NAD+ ratio and mitochondrial complex I activity to impair mitochondrial respiration in IDHWT GBM cells.

Distinct to BCAT1 KO (Supplementary Fig. S7C), BCAT2 KO significantly decreased the ratio of NAD+/NADH in U251MG cells (Supplementary Fig. S8A). Consistently, BCAT2 KO modestly increased the basal OCR and intracellular ATP levels, although it had no effect on maximal respiratory capacity and mitochondrial membrane potential in U251MG cells (Supplementary Fig. S8B–S8E). These results indicate that BCAT1 and BCAT2 have a distinct role in mitochondrial respiration in IDHWT GBM cells.

Remarkably, cell-permeable AKG synergistically increased the ratio of NAD+/NADH in BCAT1 KO#1 U87MG cells and BCAT1 KO#2 and KO#3 U251MG cells (Fig. 4,A and B). Supplementation with KIC (1 mmol/L) significantly reversed a synergistic increase in NAD+/NADH ratio conferred by BCAT1 KO/AKG in U87MG cells (Fig. 4A). Similar results were also observed in cell-permeable AKG-treated BCAT1 KO#3 U251MG cells when expression of BCAT1 was rescued (Fig. 4B). To determine whether NAD+/NADH imbalance causes BCAT1i/AKG-induced IDHWT GBM cell death, we treated parental and BCAT1 KO#2 U251MG cells with 1 mmol/L NADH in the presence or absence of cell-permeable AKG as previous studies showed that NADH can be transported into the cell (22). Supplementation with NADH partially but significantly rescued NAD+/NADH homeostasis in cell-permeable AKG-treated BCAT1 KO#2 U251MG cells (Fig. 4C). Under this condition, survival of BCAT1 KO#2 U251MG cells was restored (Fig. 4D). In contrast, NAD+ precursors nicotinamide (1 mmol/L) and NMN (1 mmol/L) failed to restore growth of BCAT1 KO#2 U251MG cells under conditions of cell-permeable AKG treatment (Supplementary Fig. S9A–S9D). We further found that NMN treatment did not replenish intracellular NADH levels in BCAT1 KO#3 U251MG cells, although it increased intracellular NAD+ levels in parental and BCAT1 KO#3 U251MG cells as well as intracellular NADH levels in U251MG cells (Supplementary Fig. S9E and S9F). Collectively, these results indicate that blockade of NADH regeneration from NAD+ during BCKA oxidation sensitizes IDHWT GBM cells to AKG treatment in vitro.

In line with the altered NAD+/NADH ratio, the OCR was significantly attenuated by cell-permeable AKG treatment in BCAT1 KO#3 U251MG cells, which was much more severe than that in parental U251MG cells treated with cell-permeable AKG or BCAT1 KO#3 U251MG cells treated with vehicle (Fig. 4E). Consistently, ATP production was synergistically decreased by cell-permeable AKG treatment in BCAT1 KO#2 U251MG cells, but not in BCAT1 rescued cells (Fig. 4F). Supplementation with KIC (1 mmol/L) also restored ATP production in cell-permeable AKG-treated BCAT1 KO U251MG or U87MG cells (Fig. 4G; Supplementary Fig. S10A). Notably, mitochondria became massive swollen 24 hours after cell-permeable AKG treatment in BCAT1 KO#3 U251MG cells and BCAT1 KO#2 U87MG cells (Fig. 4H; Supplementary Fig. S10B). In contrast, BCAT1 KO or cell-permeable AKG alone had little effect on mitochondrial morphology in IDHWT GBM cells (Fig. 4H; Supplementary Fig. S10B). Reexpression of BCAT1 rescued mitochondrial morphologic changes caused by BCAT1 KO and AKG combination (Fig. 4H). A similar rescued effect was also observed when KIC (1 mmol/L) was added into BCAT1 KO#3 U251MG cells or BCAT1 KO#2 U87MG cells treated with cell-permeable AKG (Fig. 4I; Supplementary Fig. S10B). Electron microscope images further confirmed mitochondrial swelling and loss of cristae in BCAT1 KO#3 U251MG cells treated with cell-permeable AKG, which were partially rescued by reexpression of BCAT1 or supplementation with KIC (Supplementary Fig. S10C). Mitochondrial swelling is an indicator of opening of the mitochondrial permeability transition pore and a hallmark of mitochondrial dysfunction (23). Thus, altered mitochondrial morphology confirms disruption of mitochondrial respiration and ATP production by BCAT1 KO/AKG combination in IDHWT GBM cells. Together, these results support the notion that blockade of NADH regeneration by BCAT1i/AKG combination causes death of IDHWT GBM cells possibly through mitochondrial dysfunction and energy depletion.

KIC partially prevents BCAT1i/AKG-induced nucleotide depletion

Our results about selective killing of BCAT1 KO but not BCKDHA KO U251MG cells by cell-permeable AKG (Fig. 1D; Supplementary Fig. S3G) suggest that metabolic mechanisms other than NADH depletion may also contribute to BCAT1i/AKG-induced IDHWT GBM cell death. To investigate these underlying mechanisms in greater detail, we conducted the metabolomics profiling by LC/MS in parental and BCAT1 KO#2 U251MG cells treated with or without cell-permeable AKG for 24 hours. Consistent with biochemical analysis (Supplementary Fig. S7A–S7C), NAD+/NADH was increased in BCAT1 KO#2 U251MG cells (Fig. 5A), which validated our assay. BCAT1 KO upregulated 10 intracellular metabolites but downregulated 33 intracellular metabolites in U251MG cells (P < 0.05 and 1.3-fold change as cutoffs; Fig. 5A). Of 33 downregulated metabolites, 17 are nucleotides and nucleotide derivatives and precursors (Fig. 5A). The levels of intracellular aspartate, which is involved in de novo synthesis of purines and pyrimidines, were also reduced in BCAT1 KO#2 U251MG cells (Fig. 5A). Furthermore, 6-phosphate-D-gluconate and 3-phospho-serine from the respective pentose phosphate pathway and serine-glycine biosynthesis pathway that control the flux from glucose towards purine biosynthesis were downregulated in BCAT1 KO#2 U251MG cells (Fig. 5A). Together, these results indicate that BCAT1 controls nucleotide biosynthesis in IDHWT GBM cells. Treatment of cell-permeable AKG upregulated 16 and 29 metabolic pathways but downregulated 34 and 19 metabolic pathways (FDR < 0.05) in parental and BCAT1 KO#2 U251MG cells, respectively (Fig. 5,B and C). Notably, in addition to NAD+/NADH, nucleotides were synergistically reduced in BCAT1 KO#2 U251MG cells treated with cell-permeable AKG (Fig. 5A). Purine metabolism was the sole pathway downregulated by BCAT1 KO, AKG, and both (Fig. 5B; Supplementary Table S3). No upregulated metabolic pathways were overlapped among these groups (Fig. 5C; Supplementary Table S4). Importantly, supplementation with KIC (1 mmol/L) partially rescued ATP, GTP, and CTP in AKG-treated BCAT1 KO#2 and KO#3 U251MG cells (Fig. 5D). Supplementation with nucleosides (30 μmol/L), however, failed to rescue BCAT1 KO#2 and KO#3 U251MG cells after cell-permeable AKG treatment (Supplementary Fig. S9G and S9H). Nucleotide is one of the fundamental building blocks required for cancer cell proliferation. Thus, these results suggest that nucleotides are depleted by BCAT1i/AKG combination, which may be one of mechanisms underlying BCAT1i/AKG-induced IDHWT GBM cell death.

KIC partially prevents BCAT1i/AKG-induced mTORC1 inactivation and protein depletion

Energy depletion triggers AMP-dependent protein kinase (AMPK) activation to block mTORC1 activity in the cell (12). We next studied the role of BCAT1i and/or AKG in the AMPK-mTORC1 signaling pathway in IDHWT GBM cells. Our metabolomics study showed that the AMP/ATP ratio was modestly increased by BCAT1 KO#2 (Fig. 6A). The levels of phosphorylated AMPK, but not total AMPK, were increased in BCAT1 KO#2 and KO#3 U251MG cells (Supplementary Fig. S11A), indicating that BCAT1 KO leads to AMPK activation in IDHWT GBM cells. In contrast, we found that BCAT1 KO caused mTORC1 inactivation as the putative mTORC1 substrate p70S6 kinase (p70S6K) was less phosphorylated at Threonine 389 in three individual BCAT1 KO U251MG cell lines (Supplementary Fig. S11B, left). Similar results were also observed in U87MG cells (Supplementary Fig. S11B, right). Expression of FLAG-BCAT1 in BCAT1 KO#2 U251MG cells restored phosphorylation of p70S6K (Supplementary Fig. S11C), indicating specific regulation of mTORC1 by BCAT1 in IDHWT GBM cells. In line with genetic KO of BCAT1 (Supplementary Fig. S11B and S11C), treatment of gabapentin increased AMPK phosphorylation but decreased p70S6K phosphorylation in U251MG and U87MG cells (Supplementary Fig. S11D). Previous studies showed that leucine and glutamate activate mTORC1 in cancer cells (24, 25). Supplementation with leucine, glutamate, or both in BCAT1 KO#2 and KO#3 U87MG cells failed to rescue mTORC1 activity (Supplementary Fig. S11E). Together, these findings indicate that BCAT1 inhibits AMPK activation, leading to mTORC1 activation in IDHWT GBM cells.

Similar to BCAT1 loss, the treatment of cell-permeable AKG modestly increased the AMP/ATP ratio and inhibited mTORC1 activity in U251MG cells, which were synergistically augmented in BCAT1 KO#2 and KO#3 U251MG cells (Fig. 6,A and B). The loss of mTORC1 activity was partially restored by supplementation with KIC (1 mmol/L) in U251MG cells (Fig. 6C), suggesting that BCKA depletion promotes the synergistic effect of BCAT1i/AKG combination on mTORC1 activation, similar to its effect on NAD+/NADH ratio, ATP, and nucleotides (Figs. 4,A, G, and 5D). In line with altered phospho-p70S6K, global protein synthesis was synergistically inhibited in BCAT1 KO#2 and KO#3 U251MG cells treated with cell-permeable AKG (Fig. 6D). Similar effects were also observed when U251MG cells were cotreated with gabapentin and cell-permeable AKG (Fig. 6E). Again, synergistic inhibition of protein synthesis by BCAT1 KO/AKG combination was partially rescued by supplementation with KIC (1 mmol/L) in U251MG, U87MG, and patient-derived C116 cells (Fig. 6,D and F; Supplementary Fig. S11F). Finally, we found that mTORC1 inhibitor rapamycin (1 nmol/L) phenocopied cell-permeable AKG but exhibited a less synergistic effect on death of BCAT1 KO#2 and KO#3 U251MG cells (Fig. 6G). Collectively, these results suggest that the AMPK-mTORC1-protein synthesis axis is synergistically inhibited by BCAT1i/AKG combination partially contributing to IDHWT GBM cell death.

In the current study, we defined BCAT1i/AKG as an innovative metabolic synthetic lethal approach to treat IDHWT GBM and revealed that multiple metabolic events including mitochondrial dysfunction, mTORC1 inactivation, and depletion of ATP, nucleotides, and proteins may contribute to BCAT1i/AKG-induced synthetic lethality. The intracellular BCKAs are the key determinants of BCAT1i/AKG-induced IDHWT GBM cell death.

We showed that BCAT1 is highly expressed and controls metabolic rewiring of BCAA metabolism toward mitochondrial respiration in IDHWT GBM cells, whereas BCAT2 has a minimal effect on oxidative phosphorylation, indicating that BCAT1 is an important regulator of the oxidative phosphorylation phenotype in IDHWT GBM cells. Previous studies elucidated a critical role of BCAT2 in oxidative phosphorylation in pancreatic cancer cells (7, 8). Thus, our results and others reveal a context-dependent role of BCAT1 versus BCAT2 in mitochondrial metabolism in human cancers. Our studies suggest that NAD+/NADH homeostasis during BCKA oxidation regulates oxidative phosphorylation in IDHWT GBM cells. We also found that KIC but not Leu or Glu is required for mTORC1 activation in IDHWT GBM cells. The final metabolites in the BCAA metabolic pathway are unlikely to contribute to carbons in cancer cells (6–8). Collectively, these findings indicate that the intermediates rather than the final metabolites generated from BCAA metabolism contribute to cellular building blocks and also control the cell survival signaling pathways, leading to cancer cell proliferation.

Because of Otto Warburg's seminal work in 1920s, aerobic glycolysis was thought to play a dominant role in cancer cell proliferation (26). Our studies reveal that mitochondrial dysfunction by BCAT1i/AKG combination leads to ATP depletion in IDHWT GBM cells, which further blocks the AMPK-mTORC1-nucleotide/protein signaling pathway. Interestingly, genetic KO of BCAT2 failed to sensitize AKG-induced IDHWT GBM cell death. BCAT1 and BCAT2 have a distinct role in oxidative phosphorylation in IDHWT GBM cells, highlighting the key contribution of mitochondrial dysfunction to BCAT1i/AKG-induced IDHWT GBM cell death. The distinct biochemical property of BCAT1 and BCAT2 in BCAA metabolism was also reported in pancreatic tumors, where BCAT1 controls BCAA catabolism to produce BCKAs in cancer-associated fibroblasts, while BCAT2 converts BCKAs secreted from cancer-associated fibroblasts into BCAAs in pancreatic cancer cells (8). Such an interactive function between BCAT1 and BCAT2 in mitochondrial metabolism remains unknown in IDHWT GBM tumors and needs to be defined in future. A recent report showed that 36% of IDHWT GBM tumors heavily rely on oxidative phosphorylation for their growth (18). Thus, targeting mitochondria by BCAT1i/AKG combination will have a strong clinical significance in patients with IDHWT GBM. This notion is also supported from studies in lung cancer, leukemia, and lymphoma that mitochondrion is an essential cellular organelle for cancer cell survival and proliferation and can be a therapeutic target to treat cancers (27–29).

NADH is an electron donor in the mitochondrial complex I where it transfers electrons after oxidizing into NAD+ to initiate the mitochondrial respiratory chain. We showed that BCAT1 loss increases the NAD+/NADH ratio but blocks oxidative phosphorylation in IDHWT GBM cells, which is synergized by AKG treatment. The NAD+/NADH ratio and ATP production can be partially reversed by supplementation with KIC, suggesting that NADH regeneration from NAD+ during KIC oxidation in mitochondria is crucial for oxidative phosphorylation. In agreement, we showed that BCKDHA loss robustly inhibits oxidative phosphorylation in U251MG cells. However, AKG fails to kill BCKDHA KO U251MG cells, which is most likely due to that accumulated BCKAs maintains mTORC1 activity in these KO cells. AKG can be converted to L-2-HG by lactate dehydrogenase A and malate dehydrogenases and, meanwhile, NADH is oxidized into NAD+ (30, 31). Our metabolomics study showed that 2-HG is dramatically increased but lactate is reduced in U251MG cells after AKG treatment. These results suggest that AKG may hijack lactate dehydrogenase A for its metabolism into L-2-HG along with NADH oxidation to NAD+, which leads to increase in NAD+/NADH ratio in IDHWT GBM cells. This hypothesis requires future investigation. Nevertheless, our studies reveal a molecular mechanism of depletion of the NADH pool inhibiting oxidative phosphorylation in IDHWT GBM. NAD+ and NADH are critical for the function of many cellular enzymes in human health and diseases. Previous studies have shown that NAD+ depletion is associated with neurodegeneration, heart diseases, and inflammation (32). Thus, it will be important to understand how cellular NAD+ and NADH levels are precisely manipulated to develop a potent and safe therapeutic window for IDHWT GBM treatment.

mTORC1 is a well-known regulator of synthesis of nucleotides and proteins, which are essential for cell survival and proliferation (33, 34). We showed that BCAT1 loss causes inhibition of mTORC1, which is further augmented by AKG treatment. The synergistic inhibition of mTORC1 is largely due to ATP and BCKA depletion by BCAT1i/AKG combination in IDHWT GBM cells. Leucine is a known potent activator of mTORC1 (24). Our current findings about mTORC1 activation by BCAA-derived KIC reveal another layer of regulatory mechanism of mTORC1 activation. In addition, AKG itself can inhibit mTORC1 in IDHWT GBM cells, which is consistent with a previous report showing that AKG blocks the ATP synthase activity in the mitochondrial complex V to inhibit mTOR (35). Although BCAT1 controls nitrogen donation from BCAAs for nucleotide biosynthesis (6, 7), severe inactivation of mTORC1 by BCAT1i/AKG combination at least partially leads to nucleotide depletion in IDHWT GBM cells. Therefore, multiple layers of mechanisms contribute to BCAT1i/AKG-induced disruption of cellular building blocks, leading to IDHWT GBM cell death (Fig. 6H). It is also worth noting that a mTORC1 inhibitor rapamycin has a less synthetic lethal effect than AKG, which further supports the multifunctional role of AKG in metabolic synthetic lethality in IDHWT GBM. To date, mTOR inhibitors do not show any clinical benefits in patients with GBM (36). Our findings strongly argue the necessity of targeting multiple metabolic pathways to treat IDHWT GBM. Further studies are needed to investigate whether alterations in other metabolic pathways we identified in metabolomics profiling studies also contribute to BCAT1i/AKG-induced IDHWT GBM cell death.

The synthetic lethal role of AKG in combination with BCAT1i in IDHWT GBM is unexpected since AKG is a glutamate-derived metabolite involved in energy production in the cell. Accumulating studies have uncovered several noncanonical functions of AKG in hypoxia signaling, gene regulation, cell death, cell differentiation, and immune response (37–41). The metabolic synthetic lethality is emerging in the field of cancer research given that metabolic alterations are the hallmarks of human cancers (42). Recently, several groups reported a similar role of AKG in vulnerability of cancer cells with altered cellular metabolism (38–40). Our work and others showing diverse mechanisms of AKG-mediated cancer cell death suggest a therapeutic potential of AKG in a broad range of human cancers.

In conclusion, we identified BCAT1i/AKG combination as a new synthetic lethal approach in IDHWT GBM. BCAT1i/AKG combination causes mitochondrial dysfunction and depletes the fundamental cellular building blocks including ATP, nucleotides, and proteins, leading to death of IDHWT GBM cells (Fig. 6H). Gabapentin is an FDA-approved antiepileptic drug. Our studies reveal its new potential application for a combination therapy with AKG in combating IDHWT GBM. This combined approach shows the killing effect specifically on IDHWT GBM cells but not normal astrocytes, suggesting that BCAT1i/AKG combination is a nontoxic therapeutic strategy that can be translated into the clinical investigation in patients with IDHWT GBM.

R.J. DeBerardinis reports personal fees from Agios Pharmaceuticals, Atavistik Bio, Vida Ventures, and Droia Ventures outside the submitted work. W. Luo reports grants from CPRIT, NIH, and Welch Foundation during the conduct of the study. No disclosures were reported by the other authors.

B. Zhang: Formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. H. Peng: Formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. M. Zhou: Formal analysis, investigation, methodology, writing–original draft, writing–review and editing. L. Bao: Formal analysis, investigation, writing–review and editing. C. Wang: Formal analysis, investigation, writing–review and editing. F. Cai: Formal analysis, investigation, writing–review and editing. H. Zhang: Formal analysis, investigation, writing–review and editing. J.E. Wang: Investigation, writing–review and editing. Y. Niu: Investigation, writing–review and editing. Y. Chen: Investigation, writing–review and editing. Y. Wang: Investigation, writing–review and editing. K.J. Hatanpaa: Resources, writing–review and editing. J.A. Copland: Resources, writing–review and editing. R.J. DeBerardinis: Formal analysis, writing–review and editing. Y. Wang: Conceptualization, formal analysis, supervision, funding acquisition, visualization, writing–original draft, project administration, writing–review and editing. W. Luo: Conceptualization, formal analysis, supervision, funding acquisition, visualization, writing–original draft, project administration, writing–review and editing.

The authors thank the UTSW Cancer Center Tissue Resource for assistance in IHC, which was supported by NCI Cancer Center Grant P30CA142543, the UTSW Electron Microscopy Core for assistance in sample preparation, which was supported by NIH shared instrumentation award 1S10OD021685-01A1, and the UTSW Metabolomics Core for metabolite analysis. They also thank the brain tumor PDX national resource at Mayo Clinic for providing IDHWT GBM PDX tumor. This work was supported by grants from the CPRIT (RR140036, RP220178), the NIH (R01CA222393), and the Welch Foundation (I-1903) to W. Luo, and the NIH (R35GM124693 and R01AG066166) to Y. Wang. W. Luo is a CPRIT Scholar in Cancer Research.

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

1.
Ohgaki
H
,
Kleihues
P
.
The definition of primary and secondary glioblastoma
.
Clin Cancer Res
2013
;
19
:
764
72
.
2.
Setton
J
,
Zinda
M
,
Riaz
N
,
Durocher
D
,
Zimmermann
M
,
Koehler
M
, et al
.
Synthetic lethality in cancer therapeutics: the next generation
.
Cancer Discov
2021
;
11
:
1626
35
.
3.
Karpel-Massler
G
,
Ishida
CT
,
Bianchetti
E
,
Zhang
Y
,
Shu
C
,
Tsujiuchi
T
, et al
.
Induction of synthetic lethality in IDH1-mutated gliomas through inhibition of Bcl-xL
.
Nat Commun
2017
;
8
:
1067
.
4.
McBrayer
SK
,
Mayers
JR
,
DiNatale
GJ
,
Shi
DD
,
Khanal
J
,
Chakraborty
AA
, et al
.
Transaminase inhibition by 2-hydroxyglutarate impairs glutamate biosynthesis and redox homeostasis in glioma
.
Cell
2018
;
175
:
101
16
.
5.
Peng
H
,
Wang
Y
,
Luo
W
.
Multifaceted role of branched-chain amino acid metabolism in cancer
.
Oncogene
2020
;
39
:
6747
56
.
6.
Mayers
JR
,
Torrence
ME
,
Danai
LV
,
Papagiannakopoulos
T
,
Davidson
SM
,
Bauer
MR
, et al
.
Tissue of origin dictates branched-chain amino acid metabolism in mutant Kras-driven cancers
.
Science
2016
;
353
:
1161
5
.
7.
Li
JT
,
Yin
M
,
Wang
D
,
Wang
J
,
Lei
MZ
,
Zhang
Y
, et al
.
BCAT2-mediated BCAA catabolism is critical for development of pancreatic ductal adenocarcinoma
.
Nat Cell Biol
2020
;
22
:
167
74
.
8.
Zhu
Z
,
Achreja
A
,
Meurs
N
,
Animasahun
O
,
Owen
S
,
Mittal
A
, et al
.
Tumour-reprogrammed stromal BCAT1 fuels branched-chain ketoacid dependency in stromal-rich PDAC tumours
.
Nat Metab
2020
;
2
:
775
92
.
9.
Gu
Z
,
Liu
Y
,
Cai
F
,
Patrick
M
,
Zmajkovic
J
,
Cao
H
, et al
.
Loss of EZH2 reprograms BCAA metabolism to drive leukemic transformation
.
Cancer Discov
2019
;
9
:
1228
47
.
10.
Hattori
A
,
Tsunoda
M
,
Konuma
T
,
Kobayashi
M
,
Nagy
T
,
Glushka
J
, et al
.
Cancer progression by reprogrammed BCAA metabolism in myeloid leukaemia
.
Nature
2017
;
545
:
500
4
.
11.
Ericksen
RE
,
Lim
SL
,
McDonnell
E
,
Shuen
WH
,
Vadiveloo
M
,
White
PJ
, et al
.
Loss of BCAA catabolism during carcinogenesis enhances mTORC1 activity and promotes tumor development and progression
.
Cell Metab
2019
;
29
:
1151
65
.
12.
Valvezan
AJ
,
Manning
BD
.
Molecular logic of mTORC1 signalling as a metabolic rheostat
.
Nat Metab
2019
;
1
:
321
33
.
13.
Zhang
B
,
Chen
Y
,
Shi
X
,
Zhou
M
,
Bao
L
,
Hatanpaa
KJ
, et al
.
Regulation of branched-chain amino acid metabolism by hypoxia-inducible factor in glioblastoma
.
Cell Mol Life Sci
2021
;
78
:
195
206
.
14.
Tonjes
M
,
Barbus
S
,
Park
YJ
,
Wang
W
,
Schlotter
M
,
Lindroth
AM
, et al
.
BCAT1 promotes cell proliferation through amino acid catabolism in gliomas carrying wild-type IDH1
.
Nat Med
2013
;
19
:
901
8
.
15.
Raffel
S
,
Falcone
M
,
Kneisel
N
,
Hansson
J
,
Wang
W
,
Lutz
C
, et al
.
BCAT1 restricts alphaKG levels in AML stem cells leading to IDHmut-like DNA hypermethylation
.
Nature
2017
;
551
:
384
8
.
16.
Kirby
DM
,
Thorburn
DR
,
Turnbull
DM
,
Taylor
RW
.
Biochemical assays of respiratory chain complex activity
.
Methods Cell Biol
2007
;
80
:
93
119
.
17.
Chen
Y
,
Zhang
B
,
Bao
L
,
Jin
L
,
Yang
M
,
Peng
Y
, et al
.
ZMYND8 acetylation mediates HIF-dependent breast cancer progression and metastasis
.
J Clin Invest
2018
;
128
:
1937
55
.
18.
Garofano
L
,
Migliozzi
S
,
Oh
YT
,
D'Angelo
F
,
Najac
RD
,
Ko
A
, et al
.
Pathway-based classification of glioblastoma uncovers a mitochondrial subtype with therapeutic vulnerabilities
.
Nat Cancer
2021
;
2
:
141
56
.
19.
Cheng
X
,
Geng
F
,
Pan
M
,
Wu
X
,
Zhong
Y
,
Wang
C
, et al
.
Targeting DGAT1 ameliorates glioblastoma by increasing fat catabolism and oxidative stress
.
Cell Metab
2020
;
32
:
229
42
.
20.
Hou
P
,
Kuo
CY
,
Cheng
CT
,
Liou
JP
,
Ann
DK
,
Chen
Q
.
Intermediary metabolite precursor dimethyl-2-ketoglutarate stabilizes hypoxia-inducible factor-1α by inhibiting prolyl-4-hydroxylase PHD2
.
PLoS One
2014
;
9
:
e113865
.
21.
Xiao
W
,
Wang
RS
,
Handy
DE
,
Loscalzo
J
.
NAD(H) and NADP(H) redox couples and cellular energy metabolism
.
Antioxid Redox Signal
2018
;
28
:
251
72
.
22.
Lu
H
,
Burns
D
,
Garnier
P
,
Wei
G
,
Zhu
K
,
Ying
W
.
P2×7 receptors mediate NADH transport across the plasma membranes of astrocytes
.
Biochem Biophys Res Commun
2007
;
362
:
946
50
.
23.
Zoratti
M
,
Szabo
I
.
The mitochondrial permeability transition
.
Biochim Biophys Acta
1995
;
1241
:
139
76
.
24.
Jewell
JL
,
Kim
YC
,
Russell
RC
,
Yu
FX
,
Park
HW
,
Plouffe
SW
, et al
.
Differential regulation of mTORC1 by leucine and glutamine
.
Science
2015
;
347
:
194
8
.
25.
Dyachok
J
,
Earnest
S
,
Iturraran
EN
,
Cobb
MH
,
Ross
EM
.
Amino acids regulate mTORC1 by an obligate two-step mechanism
.
J Biol Chem
2016
;
291
:
22414
26
.
26.
Warburg
O
.
On the origin of cancer cells
.
Science
1956
;
123
:
309
14
.
27.
Deribe
YL
,
Sun
Y
,
Terranova
C
,
Khan
F
,
Martinez-Ledesma
J
,
Gay
J
, et al
.
Mutations in the SWI/SNF complex induce a targetable dependence on oxidative phosphorylation in lung cancer
.
Nat Med
2018
;
24
:
1047
57
.
28.
Molina
JR
,
Sun
Y
,
Protopopova
M
,
Gera
S
,
Bandi
M
,
Bristow
C
, et al
.
An inhibitor of oxidative phosphorylation exploits cancer vulnerability
.
Nat Med
2018
;
24
:
1036
46
.
29.
Zhang
L
,
Yao
Y
,
Zhang
S
,
Liu
Y
,
Guo
H
,
Ahmed
M
, et al
.
Metabolic reprogramming toward oxidative phosphorylation identifies a therapeutic target for mantle cell lymphoma
.
Sci Transl Med
2019
;
11
:
eaau1167
.
30.
Oldham
WM
,
Clish
CB
,
Yang
Y
,
Loscalzo
J
.
Hypoxia-mediated increases in L-2-hydroxyglutarate coordinate the metabolic response to reductive stress
.
Cell Metab
2015
;
22
:
291
303
.
31.
Intlekofer
AM
,
Dematteo
RG
,
Venneti
S
,
Finley
LW
,
Lu
C
,
Judkins
AR
, et al
.
Hypoxia induces production of L-2-hydroxyglutarate
.
Cell Metab
2015
;
22
:
304
11
.
32.
Katsyuba
E
,
Romani
M
,
Hofer
D
,
Auwerx
J
.
NAD+ homeostasis in health and disease
.
Nat Metab
2020
;
2
:
9
31
.
33.
Ben-Sahra
I
,
Howell
JJ
,
Asara
JM
,
Manning
BD
.
Stimulation of de novo pyrimidine synthesis by growth signaling through mTOR and S6K1
.
Science
2013
;
339
:
1323
8
.
34.
Ben-Sahra
I
,
Hoxhaj
G
,
Ricoult
SJH
,
Asara
JM
,
Manning
BD
.
mTORC1 induces purine synthesis through control of the mitochondrial tetrahydrofolate cycle
.
Science
2016
;
351
:
728
33
.
35.
Chin
RM
,
Fu
X
,
Pai
MY
,
Vergnes
L
,
Hwang
H
,
Deng
G
, et al
.
The metabolite alpha-ketoglutarate extends lifespan by inhibiting ATP synthase and TOR
.
Nature
2014
;
510
:
397
401
.
36.
Kreisl
TN
,
Lassman
AB
,
Mischel
PS
,
Rosen
N
,
Scher
HI
,
Teruya-Feldstein
J
, et al
.
A pilot study of everolimus and gefitinib in the treatment of recurrent glioblastoma (GBM)
.
J Neurooncol
2009
;
92
:
99
105
.
37.
Tran
TQ
,
Hanse
EA
,
Habowski
AN
,
Li
H
,
Gabra
MBI
,
Yang
Y
, et al
.
alpha-Ketoglutarate attenuates Wnt signaling and drives differentiation in colorectal cancer
.
Nat Cancer
2020
;
1
:
345
58
.
38.
Sica
V
,
Bravo-San Pedro
JM
,
Izzo
V
,
Pol
J
,
Pierredon
S
,
Enot
D
, et al
.
Lethal poisoning of cancer cells by respiratory chain inhibition plus dimethyl alpha-ketoglutarate
.
Cell Rep
2019
;
27
:
820
34
.
39.
Madala
HR
,
Helenius
IT
,
Zhou
W
,
Mills
E
,
Zhang
Y
,
Liu
Y
, et al
.
Nitrogen trapping as a therapeutic strategy in tumors with mitochondrial dysfunction
.
Cancer Res
2020
;
80
:
3492
506
.
40.
Zhang
JY
,
Zhou
B
,
Sun
RY
,
Ai
YL
,
Cheng
K
,
Li
FN
, et al
.
The metabolite alpha-KG induces GSDMC-dependent pyroptosis through death receptor 6-activated caspase-8
.
Cell Res
2021
;
31
:
980
97
.
41.
Gyanwali
B
,
Lim
ZX
,
Soh
J
,
Lim
C
,
Guan
SP
,
Goh
J
, et al
.
Alpha-Ketoglutarate dietary supplementation to improve health in humans
.
Trends Endocrinol Metab
2022
;
33
:
136
46
.
42.
Zecchini
V
,
Frezza
C
.
Metabolic synthetic lethality in cancer therapy
.
Biochim Biophys Acta Bioenerg
2017
;
1858
:
723
31
.

Supplementary data