Methionine and Kynurenine Activate Oncogenic Kinases in Glioblastoma, and Methionine Deprivation Compromises Proliferation

This article is featured in Highlights of This Issue, p. 3417

Glioblastoma, grade IV malignant glioma, is one of the most aggressive types of human cancer, with a median patient survival of 12 to 15 months (1). Despite the deadliness of this disease, much is still unknown about the metabolic pathology of glioblastoma because tumor heterogeneity and genetic alterations cause an altered metabolism and metabolic profile within the same tumor, making it difficult to correlate metabolic signatures. In recent years, metabolomic-based approaches have been recognized as an emerging tool to discover products of cellular biochemical reactions that fuel cell proliferation in a variety of malignancies (2–6). Furthermore, metabolomic profiling has led to the discovery and identification of numerous key cellular pathways (7, 8). Applications of metabolomics in clinical oncology have shown strong potential in the early detection, diagnosis, and prognosis of cancer as well as being a predictive/pharmacodynamic biomarker of drug efficacy (9, 10).

Utilizing liquid chromatography/tandem mass spectrometry (LC/MS-MS), a technique extensively used for analyzing metabolites in a variety of in vitro cell culture systems (11–15), we report the differential regulation of tryptophan and methionine and their respective metabolites kynurenine and 5′-methylthioadenosine (MTA) in glioblastoma compared with normal human astrocytes (NHA). After identifying these metabolites, the primary goal of this work was to elucidate the underlying biochemical mechanisms that determine their respective roles in promoting the tumorigenic phenotype of glioblastoma. It was reported over half a century ago that cancer cells rely on methionine for proliferation and survival (16–18). Since this time, the mechanism(s) underlying this methionine dependency have not been determined. In addition, while previous studies have shown that the deregulation of the tryptophan pathway promotes immune evasion in glioblastoma (19), there is a lack of understanding of the role of the tryptophan pathway in oncogenic activation.

In this study, we have investigated the mechanisms behind the differential regulation of these metabolites in glioblastoma by evaluating the status of key enzymes in the methionine and tryptophan metabolic pathways. We have addressed the knowledge gaps that remain in the understanding of methionine dependency and the role of the tryptophan pathway in oncogenic activation. The findings reported here advance the understanding of how the methionine and tryptophan pathways contribute to the activation of oncogenic kinases and promote tumorigenesis in glioblastoma.

This study was conducted in accordance with Ohio State University (OSU) Institutional Review Boards for IRB (2009C0065), Institutional Animal Care and Use Committee (2009A0127), and IBC (2009R0169).

The 13 primary glioblastoma cell lines used in this study were isolated from glioblastoma patient tissues and authenticated by a neuropathologist at The Ohio State University (OSU34, OSU35, OSU38, OSU53, OSU61, OSU68, OSU94, OSU96, MDNSC2, MDNSC20, MHG8, MGH74, and 146). The 4 commercially available cell lines (U118, U87, LN18, and LN229) were obtained from ATCC. Glioblastoma cells were maintained in DMEM (Life Technologies), supplemented with 10% FBS (Sigma-Aldrich), and 1% antibiotic-antimycotic (Life Technologies). NHAs were obtained from Lonza and were maintained in Clonetics AGM Astrocyte Growth Medium (Lonza). Cells were cultured at 37°C under a gas phase of 95% air and 5% CO₂. All studies were conducted within five passages. Before extracting intracellular and extracellular metabolites from NHAs for metabolic profiling, the cells were grown in the same DMEM as the glioblastoma cell lines to maintain dietary consistency. To have a methionine-free DMEM for the methionine dependence studies, we used DMEM absent of glutamine, methionine, cystine, and pyruvate (Life Technologies) and supplemented with glutamine (4 mmol/L), cystine (0.2 mmol/L), and pyruvate (1 mmol/L). Methionine (20 μmol/L) was added to the methionine-free DMEM for low methionine studies.

RNA was isolated and converted to cDNA from glioblastoma cells and NHAs using the RNeasy Mini Kit and Superscript III First-Strand synthesis. Using quantitative RT-PCR, the expression level of GAPDH, IDO1, IDO2, TDO2, MTAP, AHCY, MAT1A, MAT2A, KMO, KYNU, and LAT1 were measured using the TaqMan gene expression assay. The gene expression profile data were generated using the C1000 Thermal Cycler-CFX96 Real-Time System and analyzed using the Bio-Rad CFX Manager. GAPDH was used as the control.

All results were confirmed in at least three independent experiments, and data from one representative experiment were shown. All quantitative data are presented as mean ± SD. The statistical analysis was performed using SAS 9.2 (SAS Institute) or GraphPad Prism 5. Student t tests were used for comparisons of means of quantitative data between groups. Values of P < 0.05 were considered significant.

Detailed protocols for Western blot analysis, H&E staining, cell culture and counting, metabolite extraction, global metabolic profiling using LC-MS (Q-TOF), data analysis, MTT proliferation, and colony formation assays are provided as Supplementary Information.

The cellular metabolites in the intracellular compartment and extracellular milieu of 13 primary and 4 established glioblastoma cell lines and NHAs were analyzed using liquid chromatography (LC) coupled with quadrupole time-of-flight (Q-TOF) spectrometry in positive ion mode. The results showed a distinct difference in the total number of mapped metabolites between the intracellular compartment and extracellular milieu of the glioblastoma cell lines and NHAs (Supplementary Fig. S1A and S1B). Extrapolating the data to a principal component analysis plot, the metabolic profiles of the glioblastoma cell lines and NHAs exhibited a clear separation based on their respective molecular mass, retention time, and ion abundance profiles (Supplementary Fig. S1C and S1D). This extensive list of metabolites was condensed using multivariate statistical methods (20). As a result, using NHAs as the control, we identified four statistically significant differentially regulated metabolites (DRM) in glioblastoma cells. Kynurenine and 5′-methylthioadenosine (MTA) were identified as DRMs in the intracellular compartment and extracellular milieu, while methionine and tryptophan were discovered as DRMs in the intracellular compartment alone (Fig. 1). The total and extracted ion chromatograms for both the intracellular compartment and extracellular milieu have been provided in the Supplementary Figs. S2 and S3. In addition to retention time and fragmentation pattern, the identification of these metabolites was confirmed by overlaying the molecular masses of the metabolites fragmentation patterns with the fragmentation patterns of their respective pure synthetic compounds (Supplementary Figs. S4–S47). Total ion chromatogram and multiple reactions monitoring method (MRM) were used to quantitatively validate each identified DRM (Supplementary Figs. S8 and S9). The results confirmed the increased uptake of methionine and tryptophan in the intracellular compartment and the increased secretion of kynurenine and MTA to the extracellular milieu of glioblastoma cells (Figs. 2A–D).

Given the increased uptake of the l-type amino acids methionine and tryptophan, we investigated whether this is a result of the upregulation of the l-type amino acid transporter 1 (LAT1), which exhibits high affinity for essential amino acids. The expression levels of LAT1 (Fig. 2E) did not correlate with the uptake of methionine and tryptophan, which could be a result of not considering the uptake of other l-type amino acids. Furthermore, the results suggest that the increased uptake of methionine and tryptophan is driven by the altered regulation of other cellular processes.

The intracellular concentration of methionine in the glioblastoma cell lines ranged from 5- to 100-fold higher than NHAs (Fig. 2A). Because of the fact that methionine is an essential and multifunctional amino acid required for protein synthesis, polyamine production, DNA/RNA/protein methylation, and glutathione synthesis (21, 22), we hypothesized that glioblastoma cells are dependent on dietary methionine to support their rapid proliferation, colony formation, and survival. We measured the effect of methionine on glioblastoma cell growth and colony formation by conducting MTT and clonogenic survival assays in media with and without methionine (Figs. 3A and B). Glioblastoma cells grown in the absence of methionine showed a 40%–60% inhibition of cell proliferation and a significantly decreased colony forming ability, confirming our hypothesis that glioblastoma cells are dependent on environmental methionine for survival.

When methionine enters the methionine metabolic cycle (Supplementary Fig. S10), methionine adenosyl transferase (MAT1A) and its paralog MAT2A, catalyze the conversion of methionine to the primary methyl donor of the cell, S-adenosylmethionine (SAM). SAM is then partitioned between the de novo and salvage pathways. In the salvage pathway, SAM is utilized in polyamine synthesis, generating MTA as a byproduct (21, 23). MTA is then phosphorylated by MTA phosphorylase (MTAP) producing 5′-methylthioribose-1-phosphate (MTR1P; refs.23–25). MTR1P then undergoes further processing, which produces methionine as an endproduct. Alternatively, in the de novo pathway, SAM synthesized from assimilated methionine donates methyl groups to DNA, RNA, and protein catalyzed by methyltransferases, during its conversion to S-adenosylhomocysteine (SAH; ref.26). Adenosylhomocysteinase (AHCY) converts SAH to homocysteine, which undergoes further processing and is converted back to methionine. Importantly, the SAM:SAH ratio, also defined as cellular methylation potential, directly regulates cellular methylation and gene transcription. In addition, alterations to AHCY, an enzyme also thought to be important for transmethylation reactions (27), can affect the cellular levels of SAH and corresponding methylation index (SAM:SAH). Therefore, the regulation of the SAM:SAH ratio is of interest because aberrant DNA methylation patterns are known to promote oncogenic kinase activation, tumor suppressor inactivation, and induce widespread change in signaling cascades.

To quantitatively measure the change in the DNA methylation pattern based on changes in the level of dietary methionine, we used a LC-MS/QQQ mass spectrometer to determine the SAM and SAH expression levels in glioblastoma cell lines cultured in medium containing 200 μmol/L or 20 μmol/L of methionine (Supplementary Fig. S11). To stay within the normal physiologic conditions of human cells, we used 20 μmol/L of methionine because this is the concentration found in human blood. As a group, glioblastoma cells grown in media with a decreased methionine concentration expressed significantly decreased levels of SAM, however, showed no significant alteration in SAH expression level (Fig. 3C). This disparity suggests that the increased concentration of methionine in these glioblastoma cells gets preferentially shunted to the methionine salvage pathway for protein synthesis.

Given that decreased dietary methionine significantly altered the SAM:SAH ratio in glioblastoma cells, we hypothesized that altered expression levels of key enzymes that regulate SAM and SAH cellular concentrations, MAT1A, MAT2A, and AHCY, were responsible for the dysregulation of the SAM:SAH ratio. The results showed that MAT2A was expressed by all the glioblastoma cell lines tested, while MAT1A was expressed by only one glioblastoma cell line. Subsequently, only one-third of the glioblastoma cells lines used showed significant expression levels of AHCY (Fig. 3C). The higher expression levels of MAT2A and AHCY in MDNSC2 and MDNSC20 present a correlation with the increased productions of SAM when grown in media with low methionine levels that result in a change in methylation potential. Because of the fact that SAH is a competitive inhibitor of SAM-dependent methyltransferase reactions, one possible explanation for this phenomenon may be that the increased expression levels of AHCY decrease competitive inhibition of SAM-dependent methylation, resulting in an increased SAM level. This trend is intriguing because an increase in the cellular methylation potential that regulates the reorganization of the glioblastoma methylome may contribute to the overall aggressive tumorigenic phenotype of glioblastoma.

We went on to investigate whether the altered expression of MTAP accounts for the increased secretion of MTA to the extracellular milieu by glioblastoma cells. Using qRT-PCR, we found the MTAP expression levels in glioblastoma cells were significantly lower than the expression level in NHAs (Fig. 4A). The cell lines LN18, LN229, U87, and U118 did not show any amplification of the MTAP transcript. Genomic sequencing data available for these cell lines confirmed the loss of MTAP due to LOH at 9q21. In addition, the majority of the primary cell lines used did not express CDKN2A, which could be due to the codeletion of CDKN2A and MTAP, as they are located in close proximity. Therefore, the decreased expression or LOH of MTAP results in an intracellular accumulation of MTA, likely explaining the increased secretion of MTA to the extracellular milieu by glioblastoma cells to minimize MTA-induced cytotoxicity. In addition, this loss of MTAP expression may, at least, in part, account for the increased consumption of methionine, given this loss of expression would significantly decrease the ability of these cells to recycle methionine.

To assess the effect that a MTAP loss has on the tumor-forming ability and methionine uptake of glioblastoma cells, MTAP transgene was expressed in MTAP-deficient U87 cells (U87-MTAP-OE; Fig. 4B). To estimate the relative tumor-forming ability, NOD-SCID mice were injected intracranially with the U87 and U87-MTAP-OE cell lines (28). The mice injected with the U87-MTAP-OE cell line showed an increase in survival of approximately 50 days and a significant decrease in tumor size in relation to the mice injected with the U87 cells (Fig. 4C and D). Clonogenic assays with the U87-MTAP-OE cell line demonstrated significantly decreased colony-forming ability (Fig. 4E). The concentration of methionine in the intracellular and extracellular milieu of the U87 and U87-MTAP-OE cells varied by about 2%, indicating statistically there was no change in the uptake of methionine. However, based on the concentration of MTA, the levels of conversion of methionine to MTA and MTA secretion to the extracellular milieu (15-fold decrease) were markedly compromised in the U87-MTAP-OE cells (Fig. 4F). This suggests that the methionine dependence phenotype of MTAP-deficient U87 cells cannot be reverted by the introduction of a functional MTAP gene. Our data suggest that methionine uptake is independent of MTAP gene status and intracellular concentration of MTA, rather, MTAP overexpression deregulates other genes involved in the methionine salvage pathway that contribute to the methionine dependence phenotype of glioblastoma cells.

Methionine is an essential neutral amino acid that can readily cross the blood–brain barrier through neutral amino acid transporters and accumulate in the area of an active tumor. The uptake of methionine in a normal brain is relatively low as compared with those with gliomas, hence providing a potential advantage over 2-deoxy-2-¹⁸Fluoro-D-glucose (FDG; ref.29). Previous studies have shown specificity and sensitivity of MET-PET over FDG (30–32). Our metabolic data show that accumulation of methionine is specific to glioblastoma cells. To confirm the clinical relevance of methionine as a tracer for glioblastoma, we have determined the methionine concentration in fresh patient glioblastoma biopsy tissue (Supplementary Materials; Supplementary Table S2). This is the first study which forms the foundation and rationale to use methionine as a tracer for PET imaging glioblastoma tumors.

Under normal conditions, kynurenine formed from tryptophan is converted to quinolinic acid and NAD⁺. However, during adaptive immune response, there is an increase in proinflammatory cytokines that increase the activity of the indoleamine 2,3-dioxygenase-1/2 (IDO1/IDO2) and tryptophan 2,3-dioxygenase-2 (TDO2) enzymes (33). These enzymes are responsible for the conversion of tryptophan to kynurenine and are proven immunosuppressant agents, as kynurenine suppresses antitumor immune responses in many cancer types, including glioblastoma (34). The increased uptake of tryptophan coupled with the increased secretion of kynurenine to the extracellular milieu by glioblastoma cells discovered in our metabolic profiling suggested the activation of the IDO/TDO enzymes. To investigate this activity, we measured the mRNA and protein expression levels of IDO1, IDO2, and TDO2 in glioblastomas and NHAs (Fig. 5A–C, Supplementary Fig. S12). Our results showed a significant upregulation of TDO2 in glioblastoma cells. This is in accordance with previous reports that have shown TDO2 is primarily responsible for catabolizing tryptophan to kynurenine in gliomas (19).

Under normal physiologic conditions, tryptophan is converted to the neurotransmitter serotonin. However, under pathophysiologic conditions, the rate of conversion from tryptophan to kynurenine is increased due to the upregulation of TDO/IDO enzymes. The increase in IDO/TDO activity in glioblastoma cell lines tested suggests tryptophan is shunted away from the normal neurotransmitter (serotonin) pathway and into the kynurenine pathway. In NHAs, kynurenine is almost exclusively converted to kynurenic acid, a neuroprotective compound. Conversely, kynurenine-3-monooxygenase (KMO) catalyzes the hydroxylation of kynurenine to form 3-hydroxykynurenine. Kynureninase (KYNU), a pyridoxal-5′-phosphate–dependent enzyme, then catalyzes the cleavage of kynurenine and 3-hydroxykynurenine into anthranilic and 3-hydroxyanthranilic acids, respectively. These products are then converted into quinolinic acid, a neurotoxic NMDA receptor antagonist (Supplementary Fig. S12). Kynurenine that is not degraded by KMO or KYNU may bind to the aryl hydrocarbon receptor that recruits immunosuppressive T cells, leading to immune evasion.

We hypothesized that the increased amount of kynurenine secreted to the extracellular milieu by glioblastoma cell lines tested, was a result of a change in the expression levels of KMO and KYNU. Furthermore, qRT-PCR showed a lack of expression of these enzymes in glioblastoma cells (Fig. 5D and E). Glioblastoma cells expressing KMO or KYNU had decreased amounts of kynurenine accumulation due to further metabolic conversions downstream. Therefore, an interesting correlation exists between KMO/KYNU expression and extracellular milieu kynurenine concentrations, which suggests that the lack of KMO or KYNU expression is one of the factors influencing the rate-limiting step for the utilization of kynurenine.

To discover connections between the DRMs and key oncogenic enzymes/proteins in glioblastoma cell lines, we used Ingenuity Pathway Analysis (IPA), and identified amino acid metabolism, molecular transport at the cellular level, and nervous system development as predicted functions with the highest scores (Fig. 5F and Supplementary Fig. S13). This analysis revealed that the identified DRMs were linked to a number of key oncogenic signaling proteins including PI3K, Akt, ERK, p38-MAPK, PKC, and others, that partly mediate the highly proliferative, angiogenic, invasive, and treatment-resistant nature of glioblastoma cells.

To validate the findings from our in silico analysis, we measured the activation of oncogenic protein kinases following the exogenous addition of methionine and kynurenine to glioblastoma and NHA cells (Fig. 6A). Sharp increases in the phosphorylation of Akt, PI3K, ERK1/2, and IGF1R were observed with peak intensity ranging from 5 to 30 minutes depending on the cell line; no detectable increase in phosphorylation levels of these proteins was observed in NHAs (Supplementary Fig. S14). In addition, the spikes in the phosphorylation levels of these kinases were found to be associated with a decrease in the expression of PP2A subunits (Fig. 6B). While further investigation is required to uncover the mechanistic basis behind the connection of these DRMs to the activation of key oncogenic proteins, this discovery is intriguing, as it presents the potential for novel therapeutic targets for glioblastoma.

In glioblastoma, there is an urgent need to identify biomarkers to improve diagnosis, prognosis, and treatment efficacy. In these endeavors, applications of metabolomics hold a promising potential in complementing existing functional genomics approaches. Here, we have utilized a metabolomics-based technique and discovered four differentially regulated metabolites and accessed their involvement in promoting a tumorigenic phenotype in glioblastoma.

We have shown that glioblastoma cells are dependent on dietary methionine for proliferation, colony formation, and survival. This is in accordance with a previous report that increased levels of methionine uptake in cancer cells are driven by methionine dependency (16). In addition, two previous studies have shown methionine depletion over 10 to 12 days decreases ATP and glutathione pools, sensitizing refractory tumors to the cytotoxic anticancer drugs cisplatin, doxorubicin, carmustine, N,N′-bis(2-chloroethyl)-N-nitrosourea, and temozolomide by inducing mitotic and cell-cycle arrest, apoptotic death, and widespread necrosis in tumors (17, 18). Therefore, dietary methionine and the methionine metabolic cycle may present promising therapeutic targets for combating glioblastoma.

The increased uptake of methionine by glioblastoma cells alters the SAM:SAH ratio. This is significant because changes in this ratio alter the total amount of methylation of DNA, RNA, and protein, resulting in the reorganization of the cellular methylation pattern and regulation of gene transcription. Aberrant DNA methylation patterns, typically tumor suppressor genes, are silenced by promoter hypermethylation and oncogenes are activated by promoter hypomethylation, have been associated with a large number of human malignancies. Therefore, this change in cellular methylation may activate oncogenic kinases and change cellular regulation of gene transcription that enables increased proliferation. This result reinforces the potential use of a dietary methionine restriction strategy to improve treatment outcome.

The premier biochemical and genetic etiology of methionine dependency in cancer cells is not fully understood. While deletions, polymorphisms, and altered expression of key enzymes in the methionine metabolic cycle have been implicated, there has been no report that directly links these enzymes to methionine dependency. To this end, we observed that MTAP expression is significantly decreased or undergoes a loss of MTAP due to LOH at human chromosome 9p21 in a number of tumorigenic glioblastoma cell lines. This deletion or lower expression level of MTAP coincides with previously reported data that showed in a number of cancer types, including glioblastoma, that the MTAP gene is frequently codeleted with p16INK4a (located 100 kb telomeric at locus 9p21; refs.22, 35, 36). In MTAP-deficient U87 glioblastoma cells, the expression of MTAP transgene was sufficient to compromise the tumor-forming capacity of these cells. Importantly, this suggests that MTAP may function as an independent tumor suppressor gene. This notion is further supported by previous work demonstrating that the MTAP locus can be deleted independently of p16INK4a (37). While this indicates that MTAP rescue may provide a means of increasing the survival of glioblastoma patients, the overexpression of MTAP did not alter the methionine dependency of U87 cells. A similar observation was made in MTAP-deficient MCF-7 breast cancer cells, in which overexpression of MTAP transgene reduced the tumor growth, but failed to reverse the methionine dependency (38, 39). Thus, it appears that there are mutations in other gene(s) involved in the methionine metabolic pathway responsible for rendering cells methionine dependent in glioblastoma. Moving forward, further studies are necessary to determine the underlying molecular mechanism(s) responsible for methionine dependency in glioblastoma.

The current standard of care for newly diagnosed glioblastoma is surgical resection, followed by adjuvant chemotherapy and radiotherapy (1, 40). The prolific growth of sophisticated and accurate radiotherapy techniques such as stereotactic radiotherapy, radiosurgery, intensity-modulated radiotherapy, proton therapy, etc., would allow improved survival rates to be achieved. This increased survival benefit is possible because using high-dose irradiation on a limited target volume could eradicate tumor cells while minimizing radiation exposure to adjacent normal tissues. This study forms a foundation to incorporate MET-PET imaging into the clinic and will have significant implications in the treatment strategy for glioblastoma patients for both surgical resection and radiation treatment planning. A variety of cancer cells express immunogenic antigens, but manage to escape the immune response by employing a number of immune evading mechanisms (41). One of the immune escaping mechanisms that glioblastoma cells adapt is the conversion of tryptophan to kynurenine instead of serotonin (19, 34, 42) and this reaction is fueled by activated TDO/IDO enzymes in glioblastoma. Here, we found that the intracellular level of tryptophan was higher in glioblastoma cells relative to NHAs and that glioblastoma cells produced kynurenine primarily through the catalytic action of TDO2. The lack of KMO expression and corresponding accumulation of kynurenine in the extracellular milieu suggests this may be the rate-limiting step for the utilization of kynurenine. Previous studies have shown that kynurenine, but not other tryptophan catabolic products downstream of kynurenine, directly activate aryl hydrocarbon receptors, resulting in the activation of immunosuppressant agents (43). Taken together these results suggest that reinforcing KMO or KYNU activity may result in the utilization of kynurenine, decreasing the excess amount of kynurenine in the extracellular milieu that triggers this immune evasion. A result that could restore immune surveillance. In addition, a recent report has revealed that IFNγ stimulation significantly potentiates the expression of the kynurenine pathway enzymes IDO1, IDO2, KYNU, and KMO in cultured human glioma cells (44). It has been shown that both tryptophan starvation and tryptophan catabolites (kynurenine, 3-hydroxykynurenine, 3-hydroxyanthranilic acid, and quinolinic acid) contribute to an effector mechanism of immune tolerance (45). This also supports our data suggesting that reinstating KMO or KYNU would generate kynurenine pathway catabolites that could contribute to the alteration of the regulatory T-cell environment in glioblastoma to combat the immune response. This treatment aim involving KMO or KYNU may not only restore antitumor immunity but also may reduce the capacity for malignant cells to activate oncogenic kinases. These data provides further evidence for a potential role of the tryptophan–kynurenine metabolic pathway products as novel and attractive therapeutic targets for the treatment of glioblastoma.

Interestingly, we observed that when kynurenine was added to glioblastoma cells in vitro, it induced the activation of PI3K and Akt, which correlated with the downregulation of protein phosphatase PP2A subunits responsible for reducing the phospho-Ser/Thr levels in many signaling proteins (46, 47). Previous studies have shown that inhibition of PP2A by a variety of inhibitors enhances cell growth and proliferation, which is partially attributed to the activation of the RTK–RAS–PI3K pathway (48, 49). Furthermore, it has been shown that pharmacologic inhibition of PP2A may be a general method for enhancing the effectiveness of cancer treatments that damage DNA or disrupt components of cell replication (50). Thus, our data have unfolded a new function of kynurenine in glioblastoma cells. In addition, we discovered that exogenous methionine also induced activation of these kinases and reduced the expression of the PP2A subunits. These data show a direct link between the increased uptake of methionine and kynurenine and the upregulation of key oncogenic signaling proteins. This result further supports the notion of targeting these two metabolites for the development of new glioblastoma therapies.

In sum, we have identified a four-metabolite signature that aids in the immune evasion, activation of oncogenic kinases, proliferation, survival, and development of treatment resistance in glioblastoma cells. A schematic of the metabolic alterations that induce the passage to a more progressive cancer state driven by increased tryptophan and methionine uptake is presented in the graphical abstract (Fig. 6C). While additional investigation is required to further understand the molecular mechanisms that connect these four DRMs to driving tumorigenesis, our results present exciting novel drug targets to combat glioblastoma.

No potential conflicts of interest were disclosed.

Conception and design: K. Palanichamy, A. Chakravarti

Development of methodology: K. Palanichamy, K. Thirumoorthy, A. Chakravarti

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Palanichamy, K. Thirumoorthy, S. Kanji, N. Gordon, J.R. Jacob, K.T. Litzenberg

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Palanichamy, K. Thirumoorthy, S. Kanji, R. Singh, D. Patel, T. Lautenschlaeger, A. Ray-Chaudhury, A. Chakravarti, N. Sebastian

Writing, review, and/or revision of the manuscript: K. Palanichamy, K. Thirumoorthy S. Kanji, R. Singh, J.R. Jacob, E. Bassett, B. Ramasubramanian, T. Lautenschlaeger, A. Chakravarti

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Palanichamy, J.R. Jacob, B. Ramasubramanian, A. Chakravarti

Study supervision: K. Palanichamy, A. Chakravarti

Other (helped conduct some of the experiments): D. Patel

Other (contributed reagents/materials/analysis tools): S.M. Fischer

Other (providing funding for study): K. Palanichamy, A. Chakravarti

The authors thank Dr. Sean Lawler, Brigham and Women's Hospital, Harvard Medical School (Boston, MA) for sharing with us brain tumor tissues which were used for isolating primary cells. Cell lines MDNSC2 and MDNSC20 were provided by Dr. Howard Colman, The University of Utah School of Medicine (Salt Lake City, UT), which were grown as adherent monolayers for our study. The authors also thank Dr. S. Jaharul Haque for editing the manuscript and Ananya Kamalakannan for her help with graphical abstract.

This work was supported by NIH/NCI awards R01CA108633, RC2CA148190, The Brain Tumor Funders Collaborative Group, and The Arthur G. James Comprehensive Cancer Center.

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.