IL-1β and TNF-α induce neurotoxicity through glutamate production: a potential role for neuronal glutaminase

Authors

  • Ling Ye,

    1. Department of Biochemistry and Molecular Biology, Shanghai Jiaotong University School of Medicine, Shanghai, China
    2. Center for Translational Neurodegeneration and Regenerative Therapy, Shanghai Tenth People's Hospital, Tongji University School of Medicine, Shanghai, China
    3. Laboratory of Neuroimmunology and Regenerative Therapy, Department of Pharmacology and Experimental Neuroscience, University of Nebraska Medical Center, Omaha, Nebraska, USA
    Search for more papers by this author
  • Yunlong Huang,

    Corresponding author
    1. Laboratory of Neuroimmunology and Regenerative Therapy, Department of Pharmacology and Experimental Neuroscience, University of Nebraska Medical Center, Omaha, Nebraska, USA
    • Department of Biochemistry and Molecular Biology, Shanghai Jiaotong University School of Medicine, Shanghai, China
    Search for more papers by this author
  • Lixia Zhao,

    1. Laboratory of Neuroimmunology and Regenerative Therapy, Department of Pharmacology and Experimental Neuroscience, University of Nebraska Medical Center, Omaha, Nebraska, USA
    Search for more papers by this author
  • Yuju Li,

    1. Laboratory of Neuroimmunology and Regenerative Therapy, Department of Pharmacology and Experimental Neuroscience, University of Nebraska Medical Center, Omaha, Nebraska, USA
    Search for more papers by this author
  • Lijun Sun,

    1. Laboratory of Neuroimmunology and Regenerative Therapy, Department of Pharmacology and Experimental Neuroscience, University of Nebraska Medical Center, Omaha, Nebraska, USA
    Search for more papers by this author
  • You Zhou,

    1. Center for Biotechnology, University of Nebraska Lincoln, Lincoln, Nebraska, USA
    Search for more papers by this author
  • Guanxiang Qian,

    1. Department of Biochemistry and Molecular Biology, Shanghai Jiaotong University School of Medicine, Shanghai, China
    Search for more papers by this author
  • Jialin C. Zheng

    Corresponding author
    1. Center for Translational Neurodegeneration and Regenerative Therapy, Shanghai Tenth People's Hospital, Tongji University School of Medicine, Shanghai, China
    2. Laboratory of Neuroimmunology and Regenerative Therapy, Department of Pharmacology and Experimental Neuroscience, University of Nebraska Medical Center, Omaha, Nebraska, USA
    3. Department of Pathology and Microbiology, University of Nebraska Medical Center, Omaha, Nebraska, USA
    • Department of Biochemistry and Molecular Biology, Shanghai Jiaotong University School of Medicine, Shanghai, China
    Search for more papers by this author

Address correspondence and reprint requests to Dr Jialin C. Zheng or Dr. Yunlong Huang, Laboratory of Neuroimmunology and Regenerative Therapy, Departments of Pharmacology and Experimental Neuroscience and Pathology and Microbiology, 985930 Nebraska Medical Center, Omaha, NE 68198-5930, USA. E-mail: jzheng@unmc.edu or yhuan1@unmc.edu ; Dr. Guanxiang Qian, Department of Biochemistry and Molecular Biology, Shanghai Jiaotong University School of Medicine, Shanghai, 200025, China. E-mail: qiangx@shsmu.edu.cn

Abstract

Glutaminase 1 is the main enzyme responsible for glutamate production in mammalian cells. The roles of macrophage and microglia glutaminases in brain injury, infection, and inflammation are well documented. However, little is known about the regulation of neuronal glutaminase, despite neurons being a predominant cell type of glutaminase expression. Using primary rat and human neuronal cultures, we confirmed that interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α), two pro-inflammatory cytokines that are typically elevated in neurodegenerative disease states, induced neuronal death and apoptosis in vitro. Furthermore, both intracellular and extracellular glutamate levels were significantly elevated following IL-1β and/or TNF-α treatment. Pre-treatment with N-Methyl-d-aspartate (NMDA) receptor antagonist MK-801 blocked cytokine-induced glutamate production and alleviated the neurotoxicity, indicating that IL-1β and/or TNF-α induce neurotoxicity through glutamate. To determine the potential source of excess glutamate production in the culture during inflammation, we investigated the neuronal glutaminase and found that treatment with IL-1β or TNF-α significantly upregulated the kidney-type glutaminase (KGA), a glutaminase 1 isoform, in primary human neurons. The up-regulation of neuronal glutaminase was also demonstrated in situ in a murine model of HIV-1 encephalitis. In addition, IL-1β or TNF-α treatment increased the levels of KGA in cytosol and TNF-α specifically increased KGA levels in the extracellular fluid, away from its main residence in mitochondria. Together, these findings support neuronal glutaminase as a potential component of neurotoxicity during inflammation and that modulation of glutaminase may provide therapeutic avenues for neurodegenerative diseases.

Abbreviations used
ANOVA

analysis of variance

ELISA

enzyme-linked immunosorbent assay

GAC

glutaminase C

HPLC

high performance liquid chromatography

KGA

kidney-type glutaminase

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide

OD

optical density

RCN

rat cortical neurons

STS

staurosporine

TCA

trichloroacetic acid

Chronic inflammation and neuronal damage are key processes of neurodegenerative diseases (See reviews at Huang et al. 2005; Viviani et al. 2004). The neuroinflammation of HIV-1-associated dementia (HAD), multiple sclerosis (MS), Parkinson's diseases (PD), and Alzheimer's diseases (AD) (See review at Smith et al. 2011), is considered one of the constitutive components of the disease pathogenesis and lesion generation. Studies have suggested a close link between the inflammatory response of the injured brain and neurotoxicity (Boutin et al. 2001; Takikita et al. 2001); however, whether the inflammation is a causative factor for neuronal damage remains unclear. In neurodegenerative diseases, reactive glia shift toward a pro-inflammatory phenotype and release cytokines, chemokines, as well as potentially neurotoxic substances including excess levels of glutamate, nitric oxide, and arachidonic acid (See reviews at Zindler and Zipp 2011). Cytokines, especially IL-1β and TNF-α, are typically elevated during neurodegenerative disease states and further promote central nervous system (CNS) inflammation. The increased levels of IL-1β and TNF-α may alter the activity of neurons (Chao et al. 1995; Bender et al. 2005), and increases in IL-1β and TNF-α have been observed before neuronal death (Esser et al. 1996; Guo et al. 1998). Furthermore, prolonged exposure to these cytokines generally leads to chronic inflammation and neuronal degeneration, which culminate into devastating CNS disease.

Glutamate is the most abundant excitatory neurotransmitter in the mammalian CNS (Komuro and Rakic 1996). This neurotransmitter is important in synaptic plasticity, learning, and development under physiological conditions (LoTurco et al. 1991; McEntee and Crook 1993). However, excessive glutamate stimulation induces excitotoxicity and has been linked to the pathological process of various CNS disorders including traumatic brain injury (Rao et al. 1998), ischemia (Benveniste 2009), spinal cord injury (Xu et al. 2004), stroke (Kanellopoulos et al. 2000), AD (Zoia et al. 2005), MS (Killestein et al. 2005), and HIV-1-associated dementia (HAD) (Zhao et al. 2004). Although many residential CNS cell types have been implicated in the increase in extracellular glutamate, the potential sources of excessive glutamate during the aforementioned disease states remain elusive.

Glutaminase, an enzyme localized in the inner membrane of mitochondria, converts glutamine to glutamate. As the predominant glutamine-utilizing and glutamate-producing enzyme in neurons, this enzyme has the potential to elevate glutamate to an excessive level and cause neurotoxicity (See review at Erdmann et al. 2006). In astrocytes, de novo glutamate synthesis takes place in the cytosol via pyruvate carboxylase entry to the tricarboxylic acid cycle and the activity of aspartate amino transferase. There are two isozymes of glutaminase: kidney-type glutaminase (GLS1) and liver-type glutaminase (GLS2), of which GLS1 is highly expressed in the brain (Baglietto-Vargas et al. 2004). GLS1 has various isoforms through alternative splicing from the same locus, including glutaminase C isoform (GAC) and kidney-type glutaminase isoform (KGA). GAC shares the same functional region with KGA and possesses a unique C-terminal (Porter et al. 2002). Our previous report showed that the up-regulation of GAC plays an important role in HIV-1 infection-induced neurotoxicity (Erdmann et al. 2009; Huang et al. 2011); however, the specific function and regulation of each isoform in neurons is still unclear. Glutamine is the most abundant amino acid present in the extracellular fluid of the brain, and as a substrate for glutaminase in vivo (Holcomb et al. 2000), this glutamine may be utilized by glutaminase for the production of extracellular glutamate.

To determine the regulation of glutaminase during neuroinflammation and its functional effects on neurons, we used inflammatory cytokines to stimulate primary cultured neurons (human neurons and rat cortical neurons, RCN). Our data demonstrated that IL-1β and TNF-α induced glutaminase expression and changed subcellular localization of glutaminase from the mitochondria into the cytosol and extracellular space. The up-regulation of glutaminase is associated with increases in both intracellular and extracellular concentrations of glutamate and with cell death in neuron cultures. These data suggest that glutaminase and its product glutamate are critical factors in proinflammatory cytokine-induced neurotoxicity. Modulation of glutaminase may serve as a viable therapeutic target to control excess glutamate in neurodegenerative disorders.

Material and methods

Isolation and culture of primary human neurons and rat cortical neurons

Human fetal brain tissue (gestational age 13–16 weeks) was obtained from elective abortions in full compliance with the University of Nebraska Medical Center (UNMC) and NIH ethical guidelines. Human neurons were isolated from human fetal brain tissue as previously described (Zheng et al. 1999). Human neurons were plated on poly-d-lysine-coated 24-well plates at a density of 2 × 105 cells/well. The neuron differentiation medium used for the cultures was neurobasal media (GIBCO, Invitrogen Corp., Carlsbad, CA, USA) supplemented with 2% B27 (GIBCO, Invitrogen Corp.), 1% penicillin/streptomycin (Sigma-Aldrich, St. Louis, MO, USA), and 0.5 mM glutamine. Neurons were deemed mature and used for experiments 14 days after initiation of in vitro cultures. Typically, 70% of the human neuronal-enriched preparations were microtubule associated protein-2 (MAP-2) immunopositive.

Primary rat cortical neurons (RCN) were prepared from cortices of embryonic day 17–18 (E17–18) Sprague-Dawley rat fetuses as previously described (Zheng et al. 2001). Briefly, the cortex was dissected and individual cells were mechanically dissociated in Neurobasal medium (GIBCO, Invitrogen Corp.) and filtered through 70 μm sterile nylon. The cells were then differentiated in the aforementioned neuron differentiation medium. Cultured neurons were assumed to be mature 7–12 days after plating. All experiments were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Nebraska Medical Center. Typically, 90% of the rat neuronal-enriched preparations were MAP-2 immunopositive.

In situ TUNEL assay

A commercial available terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay kit (In Situ Cell Death Detection kit; Roche, Manheim, Germany) was used to determine the apoptosis of human neurons following treatment with inflammatory cytokines. The percentage of TUNEL-positive cells over the total amount of neurons (4′,6-diamidino-2-phenylindole, DAPI staining of nuclei) was used to indicate the levels of neuronal apoptosis. At least ten images were acquired from each immunostained treatment group using a Nikon Eclipse E800 microscope (Nikon Instruments, Melville, NY, USA). Staurosporine (STS, 1 μM) were used as a positive control to induce neuronal apoptosis.

MTT reduction assay

Cell viability of RCN was assessed by 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reduction assay as described previously (Jiang et al. 2001; Zhao et al. 2004). Briefly, 5 × 104 cells/well were plated in poly-d-lysine-coated 96-well plates, treated with cytokines, and incubated with 10% MTT (Sigma-Aldrich) solution in Neurobasal media for 30 min at 37°C. The extent of MTT conversion into formazan by mitochondrial dehydrogenase was determined by measuring absorbance at 490 nm with a microplate reader (Bio-Rad Laboratories, Hercules, CA, USA). Glutamate (100 μM; Sigma-Aldrich) and the NMDA receptor antagonist MK801 (2 μM; Invitrogen Co.) were used to induce and block excitotoxicity, respectively.

MAP2 ELISA

Human neurons were plated on poly-d-lysine-coated 96-well plates at a density of 5 × 104 cells/well. After stimulated with IL-1β (10 ng/mL) or TNF-α (50 ng/mL) for 3 days, cells were fixed in 4% paraformaldehyde (PFA) at 20°C for 15 min. MAP-2 neuronal antigen was determined in human neurons treated with cytokines using colorimetric ELISA as described previously (Zheng et al. 2001; Constantino et al. 2011; Huang et al. 2011).

Severe combined immunodeficient HIV-1 encephalitis mice

Four-week-old male C.B.-17-SCID mice were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). Animals were maintained in sterile micro-isolator cages under pathogen-free conditions in the Comparative Medicine Animal Facilities at UNMC in accordance with ethical guidelines for care of laboratory animals set forth by the National Institutes of Health. One day after infection, HIV-1ADA-infected monocyte-derived macrophages (MDM, 5 × 105 cells in 5 μL) were injected intracranially by stereotactic methods (Persidsky et al. 1996). Replicate SCID mice received intracranial injections of phosphate-buffered saline (PBS) (sham-operated) served as controls.

Immunohistochemistry and image analysis

Human neurons were plated on poly-d-lysine-coated 15 mm coverslips in 24-well plates at a density of 2 × 105 cells/well. Cells were fixed in 4% PFA, rinsed with PBS, and then incubated overnight with mouse anti-MAP-2 or mouse anti-β-III-tubulin (Sigma-Aldrich, 1 : 400) antibody for the identification of neurons, with or without rabbit anti-KGA or GAC antibody (Dr. N. Curthoys, Colorado State University), followed by goat anti-mouse IgG Alexa Fluor 488 and goat anti-rabbit IgG Alexa Fluor 594 (Molecular Probes, Eugene, OR, USA, 1 : 400) secondary antibodies for 1 h at 20°C. All antibodies were diluted in PBS with 0.1% Triton X-100 and 2% bovine serum albumin . Cells were counter-stained with DAPI (Sigma-Aldrich). Morphological changes were visualized and captured with a Nikon Eclipse E800 microscope equipped with a digital imaging system using a 20x objective. All obtained images were imported into Image-ProPlus, version 7.0 (Media Cybernetics, Sliver Spring, MD, USA) for quantification. Ten to fifteen random fields (total 500–1000 cells per culture) of immunostained cells were manually counted.

Western blotting

Cells were rinsed twice with PBS and proteins were collected with M-PER Protein Extraction Buffer (Pierce, Rockford, IL, USA) containing a protease inhibitors cocktail (Roche Diagnostics, Indianapolis, IN, USA). Protein concentrations in the lysates were determined using a BCA Protein Assay Kit (Pierce). Proteins from lysates were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). After electrophoretic transfer to polyvinyldifluoridene (PVDF) membranes (Millipore, Billerica, MA, USA), proteins were incubated with polyclonal antibodies for GAC and KGA (Dr. N. Curthoys, Colorado State University), or β-actin (Sigma-Aldrich) overnight at 4°C followed by a horseradish peroxidase-linked secondary anti-rabbit or anti-mouse antibody (Cell Signaling Technologies, Danvers, MA, USA). Antigen–antibody complexes were visualized by Pierce ECL Western Blotting Substrate (Pierce). For data quantification, films were scanned with a CanonScan 9950F scanner; the acquired images were then analyzed on a Macintosh computer using the public domain NIH image program (developed at the US National Institutes of Health and available on the internet at http://rsb.info.nih.gov/nih-image/).

Intracellular and extracellular glutamate analysis

Intracellular glutamate detection was performed with the Amplex® Red Glutamic Acid/Glutamate Oxidase Assay Kit from Invitrogen following the manufacture's procedure. HPLC analysis for extracellular glutamate was performed as previously described (Zhao et al. 2004; Huang et al. 2011).

Statistical analyses

Data were evaluated statistically by the analysis of variance (anova), followed by a Tukey's test for multiple comparisons. Data were shown as mean ± SD, and significance was determined as < 0.05. To account for any donor-specific differences, all experiments were performed at least three times, with triplicate or quadruplicate samples in each assay.

Results

IL-1β and TNF-α induce neuronal death and apoptosis in vitro

To determine whether IL-1β or TNF-α, the main inflammatory factors increased during neurodegenerative diseases, could induce neurotoxicity, we treated human neurons with 10 ng/mL IL-1β or 50 ng/mL TNF-α for 72 h and determined the cell death in vitro. First, we described the dendritic damage, which is a characteristic process of neurotoxicity. MAP-2, a marker of neuronal dendrites and cell body, was used to label neurons for morphologic changes after cytokine stimulation. Similar to previous reports (See review at Block et al. 2007), treatment with those cytokines had a neurotoxic effect on human neurons (Fig. 1). Specifically, compared with untreated control (Fig. 1a–c), neuronal dendrites were reduced in size following treatment of IL-1β (Fig. 1d–f) or TNF-α (Fig. 1g–i). Notably, there were more nuclei than actual neurons, suggesting other cell type(s) in the neuronal culture. We have used glial fibrillary acidic protein (GFAP) to label astrocytes and found that astrocytes accounted for 20% of the cells in the culture. The culture also contained β-tubulin III-positive immature neurons (data not shown). The levels of MAP-2 were further determined by an ELISA assay that we previously described. The MAP-2 ELISA is a sensitive assay to determine the extent of neuronal injury and cell death (Zheng et al. 2001; Constantino et al. 2011; Huang et al. 2011). Indeed, MAP2 levels were significantly reduced following treatment with IL-1β (10 ng/mL) or TNF-α (50 ng/mL) for 72 h (Fig. 1j) in human neurons. Together, these data indicate that IL-1β and TNF-α induce dendritic damage and loss.

Figure 1.

IL-1β and tumor necrosis factor-α (TNF-α) both mediate neuronal damage in vitro. Human neurons were treated with IL-1β (10 ng/mL) or TNF-α (50 ng/mL) for 2 days. (a–i) Neuronal antigen was immunostained with antibody to microtubule associated protein-2 (MAP-2) (a, d, and g). DAPI was used to mark cell nuclei (b, e, and h). Panels (c, f, and i) are merged pictures of a–b, d–e, and g–h, respectively. (j) Levels of MAP-2 antigen after IL-1β, TNF-α, or staurosporine treatment were determined by ELISA. Results shown are the means ± SD. *< 0.05, **< 0.01, ***< 0.001 compared with control.

Induction of neuronal apoptosis is one of the mechanisms by which IL-1β induces neurotoxicity (Hu et al. 1997). To test whether IL-1β and TNF-α induce apoptosis in our human neuronal cultures, we treated human neurons with IL-1β or TNF-α for 24 h and detected the apoptosis by TUNEL assay (Fig. 2a, d and g). The nuclei of neurons at the TUNEL assay were marked with DAPI (Fig. 2b, e and h). The specificity of TUNEL immunoreactivity was evident since all of the TUNEL signals were overlapping with DAPI, indicating DNA fragmentation within the apoptotic nuclei (Fig. 2c, f, and i). Based on the quantification of the percentage of TUNEL-positive cells, IL-1β and TNF-α treatment groups had significantly higher numbers of TUNEL-positive cells compared with the untreated control (Fig. 2j). The results suggested that IL-1β and TNF-α induce apoptosis in human neuronal cultures. Together, our data demonstrate that IL-1β and TNF-α induce neuronal dendritic damage and apoptosis in human neurons.

Figure 2.

IL-1β and tumor necrosis factor-α (TNF-α) both mediate apoptosis in neuronal cultures. Human neurons were treated with IL-1β (10 ng/mL) or TNF-α (50 ng/mL) for 2 days. (a–i) IL-1β- or TNF-α-mediated neuronal apoptosis was determined by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assays. Apoptotic cells (TUNEL positive, a, d, and g), total cells (DAPI positive, b, e, and h), and their merged pictures (c, f, and i) were shown. (j) Quantitative assessments of neuronal apoptosis were determined through counting TUNEL positive cells (green) over total cell number (DAPI, blue). A minimum of 10 photos per treatment group were analyzed. Staurosporine-treated neurons served as positive control for apoptosis. Data represent means ± SD, and *< 0.05, ***< 0.001 in comparison to control.

IL-1β and TNF-α increase glutamate levels in neuronal cultures

Excessive levels of glutamate are well known to induce neurotoxicity (for review, see Erdmann et al. 2006). To determine whether glutamate is a potential mediator of IL-1β- or TNF-α-mediated neurotoxicity, we first measured glutamate levels in the neuronal cultures. We determined intracellular glutamate level using a commercial available glutamate detection kit. As expected, IL-1β and TNF-α induced a two- and threefold increase in intracellular glutamate, respectively, as compared to control (Fig. 3a). Since increased generation of glutamate could be an important cellular source of extracellular glutamate, we next investigated the extracellular levels of glutamate. Similar to the intracellular glutamate, the extracellular glutamate increased following the treatment with IL-1β or TNF-α (Fig. 3b). Both IL-1β (10 ng/mL) and TNF-α (50 ng/mL) also significantly reduced neuronal viability as determined by MTT assay, an effect comparable to the positive control (glutamate, 100 μM) (Fig. 3c). Interestingly, pre-treatment with MK801 significantly blocked the increase in extracellular glutamate and inflammatory cytokine-induced neurotoxicity (Fig. 3b and c). Together, these data suggest that glutamate and glutamate-mediated neurotoxicity are critical components of IL-1β- and TNF-α-induced neuronal damage.

Figure 3.

IL-1β and tumor necrosis factor-α (TNF-α) both increase intracellular and extracellular glutamate levels in neuronal cultures. Rat cortical neurons (RCN) were treated with IL-1β (10 ng/mL) or TNF-α (50 ng/mL) for 3 days. (a) Intracellular concentration of glutamate was determined by a Glutamic Acid/Glutamate Oxidase Assay Kit. **< 0.01 compared with control. (b–c) RCN were pre-treated with 2 μM MK801 (NMDA receptor antagonist) for 2 h then treated with IL-1 β or 50 ng/mL TNF-α for 3 days. The concentration of glutamate in cell-free supernatants was determined by RP-HPLC (b). RCN Cell viability was assessed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (c). Results shown are the means ± SD of triplicate samples. *< 0.05, **< 0.01, ***< 0.001 compared with control. ##< 0.01, ###< 0.001 compared with cytokine treatment alone.

IL-1β and TNF-α regulate glutaminase expression in human neurons

The increase in intracellular and extracellular glutamate following IL-1β or TNF-α treatment indicates an increase in glutaminase activity in the neuronal culture. We previously demonstrated that the increased glutamate production from HIV-1-infected macrophages and microglia was dependent on mitochondrial glutaminase (Zhao et al. 2004; Huang et al. 2011). Therefore, we further investigated whether glutaminase, particularly its two main isoforms KGA and GAC, is responsible for glutamate generation and the neurotoxicity induced by IL-1β or TNF-α. The expression of KGA and GAC in human neurons during the process of cytokine-mediated neurotoxicity was first assessed by immunocytochemistry. Neuronal cultures were labeled with KGA (Fig. 4a) and with a neuronal marker β-tubulin III (Fig. 4b). The KGA had a punctate distribution in the cytoplasm and along the dendrites, which parallels the distinct distribution pattern of the mitochondrial network (Fig. 4a). The colocalization of KGA and β-tubulin III suggested that KGA is primarily expressed by the neuronal population (Fig. 4c). After IL-1β or TNF-α treatment, KGA immunoreactivity increased in a pattern similar to mitochondrial distribution when compared with untreated control (Fig. 4d–i). In contrast, the expression of GAC in neurons decreased after IL-1β or TNF-α treatment (Fig. 4j–r). Together, the immunocytochemistry data indicate that the IL-1β and TNF-α regulate the glutaminase isoforms in neuronal cultures.

Figure 4.

IL-1β and tumor necrosis factor-α (TNF-α) both regulate glutaminase staining in neurons. Human neurons were treated with IL-1β (10 ng/mL) or TNF-α (50 ng/mL) for 24 h and then immunostained with kidney-type glutaminase (KGA) (a, d, g), GAC (j, m, p), or β-tubulin III (neuronal marker, b, e, h, k, n, q). Panels (c, f, i, l, o, and r) are merged pictures of a–b, d–e, g–h, j–k, m–n, and p–q, respectively. Images were acquired from a Bio-Rad MRC1024ES LASER scanning confocal microscope. Magnifications: (a-f). 600X. Panels are representative of three separate donors.

To further determine the regulation of glutaminase during the cytokine stimulation, we investigated the KGA and GAC mRNA and protein levels in neuronal cultures following IL-1β or TNF-α treatment. Gene expression analysis using real-time RT-PCR revealed a 2.2-fold and 1.7-fold increase of the KGA mRNA following IL-1β and TNF-α treatment, respectively, when compared to untreated human neurons (Fig. 5a). The up-regulation of KGA by inflammatory cytokines was further confirmed at the level of protein translation through western blotting. IL-1β and TNF-α each induced 5-fold increase of KGA compared with untreated control (Fig. 5c and e). The up-regulation of KGA is specific because after inflammatory cytokines stimulation, both the mRNA (Fig. 5b) and protein levels (Fig. 5d and f) of GAC had trended downward. The specific up-regulation of KGA suggests that KGA may be the isoform responsible for the IL-1β and TNF-α induced neurotoxicity.

Figure 5.

IL-1β and tumor necrosis factor-α (TNF-α) both regulate glutaminase mRNA and protein expression in neurons. Human neurons were treated with IL-1β (10 ng/mL) or TNF-α (50 ng/mL) for 2 days. (a–b) RNA was collected and expression of glutaminase isoforms kidney-type glutaminase (KGA) (a), glutaminase C (GAC) (b) were analyzed using Real-time RT PCR. Data were normalized to GAPDH and presented as fold change compared to control. * denotes p < 0.05 as compared to control. (c–f) Whole cell lysates were collected and protein levels of KGA (c) and GAC (d) were analyzed by western blot. Actin was used as the loading control. Levels of KGA (e) and GAC (f) were normalized as a ratio to actin after densitometrical quantification and presented as fold change relative to control. Results are expressed as the mean ± SEM of triplicate samples from three independent experiments. *< 0.05 compared to control.

Glutaminase is elevated in a murine model of HIV-1 encephalitis

To further determine the relevance of glutaminase during brain inflammation in vivo, we obtained an established murine model of HIV-1 encephalitis. The injection of HIV-1-infected MDM is known to elicit a sustained inflammatory response reminiscent of HIV-1 encephalitis brain tissue (Persidsky et al. 1996). Notably, astrogliosis and inflammatory cytokines TNF-α and IL-1β were evident after MDM inoculation (Persidsky et al. 1996). We have previously correlated the elevation of IL-1 β in this animal model to the dysregulation of CXCL12, a chemokine important for leukocyte and stem cell migration (Peng et al. 2006). To determine whether the inflammatory response was related to the glutaminase, we used a polyclonal antibody against glutaminase (Dr. N. Curthoys, Colorado State University) to identify the glutaminase expression in vivo (Fig. 6a and b). The inflammatory area in the HIV-1 encephalitis model was defined by the GFAP immunoreactivity, which marked astrocyte activation (Fig. 6c and d). Glutaminase expression in the inflammatory area had little colocalization with the GFAP, suggesting lower expression levels of glutaminase in astrocytes. In contrast, a majority of glutaminase immunoreactivity appeared to have morphology characteristic of neuron (Fig. 6a, b lower left inserts). Furthermore, glutaminase was co-localized with β-tubulin III, a neuronal marker, suggesting a predominant expression of glutaminase in neurons (Fig. 6e). Importantly, glutaminase expression in the neuroinflammatory area was significantly increased compared with sham-operated mice (Fig. 6f), suggesting that glutaminase may be associated with inflammatory tissue pathology.

Figure 6.

Glutaminase expression in a murine model of HIV-1 encephalitis. phosphate-buffered saline (PBS) or HIV-1ADA-infected monocyte-derived macrophages (MDM) (5 × 105 cells in 5 μL) were intracranially injected into the basal ganglia of SCID mice. After 7 days, brain tissues were collected and subjected to immunohistochemical staining. (a–d) Serial floating coronal sections from PBS (a, c) and HIV-1-infected MDM (b, d) injections were immunolabeled with antibodies to GLS 1 (red) and GFAP (green). Higher magnification GLS1 staining images for the PBS (a) and HIV-1-infected MDM (b) affected brain area were shown in the lower left inserts. (e) Brain sections were stained with neuronal marker β-tubulin III (green) and GLS1 (red). (f) GLS1 expression from panels (a and b) was quantified by determining the GLS1-positive area as a percentage of the total image area per microscopy field and calculated for a window of tissue immediately surrounding the injection site. Three sections of at least four mice were examined for each group. **p < 0.01 compared with PBS group. Scale bar represents 200 μM. Scale bar in the lower left inserts represents 25 μM. CC, corpus callosum; CPu, Caudate putamen.

IL-1β and TNF-α regulate subcelluar localization of KGA

We previously reported that glutaminase released from mitochondria contributes to the neurotoxicity of HIV-1 infected macrophages (Erdmann et al. 2009; Tian et al. 2012). To identify changes in glutaminase distribution following cytokine stimulation, particularly at locations away from its main residence in mitochondria, we used subcellular fractionation and subsequently western blotting for KGA. Mitochondrial and cytosolic fractions from control and cytokine-treated neurons were isolated and the enrichments of mitochondria and cytosol were confirmed by the blotting of voltage-dependent anion channels (VDAC), a specific outer mitochondrial membrane protein not present in the cytosolic factions (Fig. 7a). Analysis of KGA levels in the mitochondrial and cytosolic fractions revealed that the cytosolic fraction from control human neurons had detectable levels of KGA present, possibly because of the process of in vitro manipulation of the neuronal cultures. Interestingly, in the cytosolic fractions, KGA levels were increased in both IL-1β- and TNF-α-treated neurons (Fig. 7a and b), indicating that proinflammatory cytokines induce a release of KGA from mitochondria into the cytosolic compartment of human neurons.

Figure 7.

IL-1β and tumor necrosis factor-α (TNF-α) induce glutaminase release from mitochondria in vitro. Human neurons were treated with IL-1β (10 ng/mL) or TNF-α (50 ng/mL) for 3 days. (a) Cells were subjected to subcellular fractionation; the cytosolic and the mitochondrial fractions were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and blotted for immunodetection of kidney-type glutaminase (KGA). Voltage-dependent anion channels (VDAC) and β-actin were used as loading controls for the mitochondrial and cytosolic fractions, respectively. (b) Levels of KGA in cytosol and mitochondria were normalized as a ratio to VDAC or β-actin after densitometrical quantification and presented as fold change relative to the untreated control. (c) Proteins from equal volumes (30 mL) of conditioned-medium from cultures of control or cytokine-stimulated human neurons were precipitated using TCA. KGA levels in the precipitated protein were determined by western blotting. (d) Results of C were normalized with β-actin and densitometrically quantified as fold change relative to the untreated control. All the data are representative of at least three independent experiments with human neurons from three different donors.

TNF-α induces glutaminase release into the extracellular space

Glutaminase release into the extracellular fluid is one of the important mechanisms for excess glutamate production in HIV-1-infected macrophages (Zhao et al. 2004; Erdmann et al. 2009). However, whether there is glutaminase release in neuronal cultures remains unclear. We collected equal amounts (30 mL) of control and cytokines-treated neuron culture supernatants and precipitated the proteins with trichloroacetic acid (TCA). Protein pellets were re-suspended in lysis buffer, denatured, and glutaminase was detected via western blotting (Fig. 7c). Interestingly, there was no evidence of KGA release in control and IL-1β treated samples. In contrast, TNF-α treatment induced a marked increase in KGA levels in the supernatant (Fig. 7c and d). These findings demonstrated that glutaminase was specifically released to the extracellular space in human neurons following TNF-α treatment.

Discussion

One of the most important roles of glutaminase in the CNS is its contribution to excitatory neurotransmitter in neuron. Although the effects of macrophage and microglia glutaminases in brain injury, infection, and inflammation are well studied (Newcomb et al. 1997; Zhao et al. 2004; Dohmen et al. 2005; Tian et al. 2008; Erdmann et al. 2009; Huang et al. 2011), little is known about the regulation of neuronal glutaminase. In this study, we found that IL-1β and TNF-α induced neuronal death and apoptosis in vitro through a mechanism involving the increase in glutamate levels in culture supernatants (Figs 1-3). Further analysis of glutaminase isoforms revealed that IL-1β and TNF-α up-regulated KGA expression in human neurons (Figs 4, 5). Interestingly, both IL-1β and TNF-α appeared to regulate the subcelluar localization of KGA, inducing KGA into the cytosol and away from its mitochondrial residence (Fig. 7a and b). Furthermore, TNF-α induced glutaminase release into the extracellular space, which is a distinct event not seen in IL-1β-treated or control neuronal cultures (Fig. 7c and d). Together, our data suggest that IL-1β and TNF-α induce neurotoxicity through glutamate, the mechanism of which is through dysregulation and translocation of neuronal glutaminase. These findings provide insights into the critical role of excess glutamate levels and dysregulation of neuronal glutaminase in inflammation-induced neurotoxicity.

Through unique primary neuronal cultures, we have determined that IL-1β and TNF-α up-regulate neuronal KGA expression and stimulate extensive neuronal KGA release, resulting in increased glutamate production and excitotoxicity. The discovery of this cascade of neurotoxic processes initiated by cytokines is significant because it describes the inflammatory perturbation of brain homeostasis that leads to neuronal deficits. Moreover, the up-regulation of cytokines IL-1β and TNF-α is a common feature of glial cell activation in several inflammatory neurodegenerative diseases (Hu et al. 1995; Van Eldik et al. 2007). Although IL-1β and TNF-α alone or in combination with other cytokines have been shown to promote cell death and apoptosis (Chao et al. 1995; Bender et al. 2005), the molecular mechanism remains poorly defined. The identification of the neurotoxic effects of IL-1β and TNF-α through mechanisms involving glutaminase will therefore be important to the understanding of excess glutamate production during brain inflammation.

Little is known about how KGA and GAC, two isoforms of GLS1, are regulated. One recent study suggests a role for mir-23a/b and NFkB in the regulation of GAC (Gao et al. 2009). Furthermore, we have identified that type I interferon regulates GAC expression through STAT-1 binding (Zhao et al. 2012). In this study, we have demonstrated that inflammatory cytokines increase KGA isoform in neurons (Figs 4, 5). Because KGA is expressed at a higher level in cultured neurons than in glia (Data not shown), the up-regulation of KGA by inflammatory cytokines indicates neuronal KGA may be more important in cytokine-induced neurotoxicity than GAC. Our study, along with several recent studies, defines the importance of understanding the mechanism of glutaminase-mediated excess glutamate production in disease pathogenesis, including viral infection, cancer development, and neurological diseases (Maezawa and Jin 2010; Wang et al. 2010; Huang et al. 2011). However, the physiological relevance of the novel glutaminase regulation that we identified in neurons remains incompletely understood. Using immunohistochemical technique, we have demonstrated the up-regulation of glutaminase in situ in a murine model of encephalitis. The great majority of glutaminase immunoreactivity in the affected mouse brain area was found to be in β-tubulin III-positive neurons (Fig. 6). Further studies are needed to help determine the functional importance of neuronal glutaminase in vivo.

Previous studies in our group showed that in addition to glutaminase expression, glutaminase activity, or release mediated by the infective process of HIV-1 may facilitate excess generation of glutamate in the CNS (Erdmann et al. 2009; Tian et al. 2012). Using the subcellular fractionation for neurons, we found that inflammatory cytokines induced neuronal glutaminase translocation from mitochondria to cytosol (Fig. 7a and b). Interestingly, widespread neuron death could result in glutaminase release, excess glutamate production, and further neurotoxicity in vitro (Newcomb et al. 1997). We further investigated the glutaminase level in culture supernatants. After precipitating protein in culture supernatants, glutaminase was found in TNF-α-stimulated extracellular protein precipitates. However, we were unable to detect glutaminase release in the culture supernatants of IL-1β-stimulated group (Fig. 7c), suggesting that glutaminase release from cells is specific to TNF-α treatment. The specific release of KGA could make promising clinical targets to develop novel therapeutic strategy against excitotoxicity by free glutaminase. For IL-1β, since it increases KGA expression and both the intracellular and extracellular levels of glutamate (Figs 3-5), it is likely that IL-1β mediates excess glutamate production through a mechanism independent of extracellular glutaminase release.

The interaction between cytokines and the glutamine metabolism is of great importance in inflammatory brain diseases (See review at Viviani et al. 2004). Increased glutamate levels have been suggested as responsible for the neuronal death characteristic of many neurodegenerative diseases (See review at Meldrum 2000). In this study, we have demonstrated that both IL-1β and TNF-α up-regulated intracellular and extracellular glutamate levels (Fig. 3). The cell death in the neuronal cultures is likely a consequence of glutamate increase, since MK801, a NMDA receptor antagonist, blocked the glutamate production and cell death (Fig. 3). The NMDA receptor has a critical role in mediating excitotoxic neuronal cell death in a variety of neurodegenerative conditions (See review at Leist and Nicotera 1998). Recent evidence indicates that IL-1 may mediate increased neuronal excitability through direct interaction of its receptor complex with NMDA receptors and inhibition of Ca2+-induced K+ channels (Zhang et al. 2008; Gardoni et al. 2011). Whether both IL-1β and TNF-α receptors have direct interaction with NMDA receptors in the neurons remains to be determined.

In summary, we have identified neuronal glutaminase as an important player in mediating glutamate over-production during inflammatory cytokines stimulation. The potential effect of glutaminase on glutamate involves dysregulation, translocation, or release of glutaminase isoforms, which consequently induce neurotoxicity and apoptosis. Uncovering the critical role of glutaminase in inflammation-induced neurotoxicity may provide a potential target for therapeutic intervention in inflammatory brain diseases.

Acknowledgements

We acknowledge Beibei Jia for providing technical support, Dr N. Curthoys (Colorado State University) for the antibody against KGA, GAC, and GLS1, Randall J Ambroz, Kristin M. Leland Wavrin for editing the manuscript for this study. Tiffany R. Peña, Julie Ditter, Robin Taylor, Johnna Belling, Na Ly, Myhanh Che, Mary Cavell, and Emilie Scoggins provided outstanding administrative support. This study was supported in part by research grants by the National Institutes of Health: R01 NS 41858-01, R01 NS 061642-01, 3R01NS61642-2S1, R21 MH 083525-01, P01 NS043985, and P20 RR15635-01 (JZ) and National Natural Science Foundation of China (NSFC) # 81028007. The authors indicate no potential conflicts of interest.

Ancillary