Glutaminase 1 regulates the release of extracellular vesicles during neuroinflammation through key metabolic intermediate alpha-ketoglutarate

To further study the functional impacts of GLS1 on EV release, we constructed adenovirus vectors that overexpressed KGA or GAC to mimic the upregulation of these isoforms during HIV-1 infection. The extracellular levels of glutamate increased significantly in both KGA- and GAC-overexpressing HeLa cells at the multiplicities of infection (MOI) of 200 compared with those in uninfected or vector-treated cells (Additional file 1: Figure S1A). An MOI of 200 was then used for the following GLS1-overexpression experiments. The overexpression of KGA and GAC was confirmed by Western blot (Additional file 1: Figure S1B). GLS1 enzyme activity assay confirmed the increase in KGA and GAC activity in protein lysates from KGA- and GAC-overexpressing HeLa cells, suggesting that the overexpressed KGA and GAC were functional (Additional file 1: Figure S1C). Consistent with GLS1 enzyme activity, the levels of intracellular glutamate (Additional file 1: Figure S1D) and extracellular glutamate (Additional file 1: Figure S1E) were increased after KGA and GAC overexpression compared with those in the GFP and control groups.

After confirming KGA and GAC overexpression, we investigated GLS1 activities in the extracellular fluid. We found that cell-free supernatants showed increased glutamate production after the addition of glutamine, indicating that GLS1 overexpression leads to elevated levels of GLS1 activities in the extracellular fluid. L-DON, a GLS1 inhibitor, blocked the excess generation of glutamate in the supernatants, confirming that the observed GLS1 activity in the extracellular supernatants was from the protein GLS1 (Fig. 1a). To confirm that the observed GLS1 activity in the cell-free supernatants was from EVs, we isolated EVs from cell-free supernatants and tested the EVs in GLS1 activity assay. EVs that were isolated from KGA- or GAC-overexpressing HeLa cells generated significantly higher levels of GLS1 when incubated with glutamine (Fig. 1b). When L-DON was added to the enzyme reactions, the generation of glutamate was blocked, suggesting that KGA and GAC overexpression induced EV release that contains high levels of GLS1 activities (Fig. 1b). To determine the EV regulation after KGA and GAC overexpression, proteins from EVs were subjected to Western blot for EV markers, including tTG and flotillin-2. The levels of tTG and flotillin-2 did not change in the whole cell lysates but specifically increased in EVs collected from KGA- and GAC-overexpressed HeLa cells (Fig. 1c–f). The mitochondrial protein cytochrome c was absent in the EV preparation, suggesting that the preparation was not contaminated with subcellular organelles such as mitochondria. Together, these data suggest that elevation of GLS1 is sufficient to induce the extracellular release of EVs.

Our previous work demonstrated the role of GLS1 in the regulation of EV release [17]. However, the mechanism of how GLS1 regulates the release remains unclear. To determine whether glutamine metabolism is essential for EV release, we added different concentrations of glutamine to HIV-1-infected macrophages for 1 day and then isolated EVs from the cell-free supernatants. The EVs were first subjected to negative staining TEM for characterization. EVs manifested typical morphologies in TEM and had sizes ranged from 40 to 300 nm (Additional file 1: Figure S2A–D). Quantification of EVs showed a significant increase of EVs at a concentration of 1 mM and higher concentrations of glutamine did not appear to further increase EV release (Additional file 1: Figure S2E). Next, the addition of glutamine was confirmed with measurements of glutamate and glutamine in cell-free supernatants by reversed-phase high-performance liquid chromatography (RP-HPLC, Fig. 2a, b). To determine the EV regulation after glutamine treatment, we collected protein lysates from both the whole cells and EVs. The expression levels of GAC under different concentrations of glutamine did not change compared to those in control group (Fig. 2c). EV markers Alix and flotillin-2 did not change in whole cell lysates but specifically increased in glutamine treatment groups at 1 and 2 mM compared to controls (Fig. 2c–f). Glutamine treatment at 5 mM did not induce significantly higher levels of EV release compared with the control. Together, these results suggest that EV release in HIV-1-infected macrophages is dependent on glutamine metabolism.

To investigate whether EV release is dependent on glutamine in microglia, a BV2 microglial cell line was used. Different concentrations of glutamine were added to the cultures 6 h prior to lipopolysaccharide (LPS) treatment. After overnight treatment, the levels of extracellular glutamate were increased with the addition of glutamine in a dose-dependent manner (Fig. 3a, b). To determine the EV regulation, protein lysates from whole cells and EVs were subjected to Western blots. In the whole cells, GAC showed a trend of increase with the addition of glutamine concentrations, whereas EV markers remained unchanged (Fig. 3c). In EV lysates, EV markers Alix and flotillin-2 were both significantly increased with the addition of glutamine (Fig. 3d–f). Consistent with Western blot analysis, nanoparticle tracking analysis (NTA) showed that addition of glutamine significantly increased EV release in a dose-dependent manner (Fig. 3g, h). Altogether, these results demonstrated that EV release in immune-activated microglia is dependent on glutamine.

To further investigate whether GLS1-mediated glutamine metabolism is crucial for EV release in HIV-1-infected macrophages and immune-activated microglia, we used BPTES, a potent GLS1 inhibitor [18]. Human macrophages were infected with HIV-1 virus with 10 μM BPTES prior to sample collections (Fig. 4a). When BPTES was added 1 day prior to EV isolation, EV lysates showed significantly decreased levels of EV markers Alix, Flotillin-2, and tTG compared with those from HIV-1-infected macrophages (Fig. 4b–e). In contrast, when BPTES was added immediately after the HIV-1 infection, the EV regulation was not affected, indicating that BPTES affects the EV release within a limited time frame.

To investigate GLS1-mediated glutamine metabolism in immune-activated microglia, we used BPTES and CB839, both of which are potent GLS1 inhibitors [19]. BPTES or CB839 was added to BV2 cells prior to LPS treatment, and EV regulation was determined through Western blot and NTA (Fig. 5a). LPS, BPTES, and CB839 treatment did not significantly change BV2 cell viability (Additional file 1: Figure S3). In contrast, both levels of Alix and flotillin-2 were significantly increased after LPS stimulation in the EVs isolated from BV2 cells, suggesting that LPS increases EV release in BV2 cells (Fig. 5b–d). Pretreatment with BPTES and CB839 reduced the levels of Alix and flotillin-2 in LPS-activated BV2 cells (Fig. 5b–d). Consistent with Western blot data, NTA revealed that treatment with BPTES and CB839 significantly reduced the EV particles in both the LPS-activated and unactivated control BV2 cells. Collectively, these results indicated that the release of EVs in HIV-1-infected macrophages and LPS-treated microglia is dependent upon GLS1-mediated glutamine metabolism.

GLS1-mediated glutamine metabolism produces glutamate and subsequently α-ketoglutarate (α-KG). To determine whether α-KG is required for EV biogenesis and release, 1 mM of α-KG was added to LPS-stimulated BV2 cells with GLS1 inhibitors. Treatment with BPTES and CB839 significantly reduced GLS1 activity in the presence of 1 mM α-KG (Fig. 6a), confirming that they are indeed GLS1 inhibitors. In the presence of α-KG, BPTES and CB839 failed to decrease EV markers Alix and flotillin-2 in EV lysates (Fig. 6b-d). Consistent with the Western blots, NTA revealed that suppression of EV release by BPTES and CB839 was reversed by α-KG (Fig. 6e, f). These results suggest that α-KG is the critical downstream effector that is responsible for the GLS1-mediated EV release.

Our previous report has shown that the release of EV can be blocked by GW4869, a nSMase inhibitor [6, 20]. nSMase is known to catalyze the production of ceramide, which is an active component of EVs. Therefore, we hypothesized that GLS1 mediates EV release through sphingolipid metabolism. GLS1 activity remained suppressed by BPTES and CB839 in the presence of C₆ ceramide (Fig. 7a), a cell-permeable short-chain ceramide that can be supplied to cell cultures as exogenous ceramides to increase cell ceramide levels [21]. However, NTA showed that BPTES- and CB839-mediated inhibition of EV release was blocked by C₆ ceramide (Fig. 7b), indicating that the ceramide pathway of sphingolipid metabolism is involved in the GLS1-regulated release of EVs.

To determine the link between the GLS1 and the secretion of EVs in vivo, we used a GLS1 transgenic mouse model that we recently generated [17]. The transgenic mice have GLS1 isoform GAC overexpressed in all cells of the neural lineage. Positive mice were termed Nestin-GAC transgenic mice. We successfully isolated EVs from the Nestin-GAC transgenic brain tissues following the protocol from a previous publication [22]. After normalized EVs to equal amount of initial brain tissues used for EV isolation, we found that Nestin-GAC mice had significantly higher levels of EV markers Alix and Flotillin-2 in the isolated EVs compared with those from control mice, suggesting that brain-specific overexpression of GLS increases EV release in vivo (Fig. 8a–c). To further characterize the EVs isolated from in vivo tissues, Thy1-GAC and wild-type (WT) were generated by crossing Thy1-Cre mice and CAG-loxp-GAC mice, and PCR reactions were performed to confirm the genotypes of the mice. EVs were isolated and purified from the hemibrains of both WT and Thy1-GAC mice and subjected to negative staining TEM analysis. The number of EVs was significantly higher compared with those in WT control (Fig. 8d–f). Together, these data indicate that the brain-specific GAC overexpression increases EV release in vivo.

MDM were used in full compliance with the University of Nebraska Medical Center and National Institutes of Health ethical guidelines, with the Institutional Review Board (IRB) #: 162-93-FB. We have the informed written consent from all participants involved in this study. All mice were housed and bred in the Comparative Medicine Animal Facilities at the University of Nebraska Medical Center. All procedures were conducted in accordance with the protocol (11-018-04) approved by the Institutional Animal Care and Use Committee at the University of Nebraska Medical Center.

Human peripheral blood-derived mononuclear cells were isolated through leukopheresis from healthy donors. Human macrophages were differentiated in Dulbecco’s modified Eagle’s media (DMEM) (Sigma-Aldrich, St. Louis, MO) with 10% human serum, 50 μg/ml gentamycin, 10 μg/ml ciprofloxacin (Sigma), and 1000 U/ml recombinant human macrophage colony-stimulating factor (MCSF) for 7 days. Human fetal microglial cells were obtained from fetal brain tissue-derived microglia-astrocytes mixed cultures as previously described [46]. The HIV-1_ADA strain was used to infect the macrophages and microglia at a multiplicity of infection (MOI) of 0.1 and 0.5, respectively. After 24 h, the culture medium was changed to remove any remnant virus. Seven days after HIV-1-infection, the culture medium was changed to glutamine-free neurobasal medium for 24 h, and supernatants were collected for subsequent HPLC or Western blot analysis. HeLa and BV₂ cell lines were obtained from ATCC, and both cell lines were grown in DMEM with 10% fetal bovine serum and antibiotics. LPS was used to immune activate BV₂ cells for 24 h, and supernatants were collected for HPLC and Western blot analysis. Bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl)ethyl sulfide (BPTES) and CB839 (generous gifts provided by Dr. Takashi Tsukamoto from John Hopkins University and later ordered from Millipore with catalog numbers 530030 and 533717, respectively) were used in HIV-1-infected macrophages or LPS-treated microglia prior to EV isolation. α-Ketoglutarate (Sigma; 349631) and C₆ ceramide (Sigma; H6524) were also used to manipulate the metabolic intermediates in HIV-1-infected macrophages or LPS-treated microglia. All experiments involving human cell samples are approved by the Institutional Review Board at the University of Nebraska Medical Center.

Replication-defective adenovirus vectors expressing human KGA and GAC were generated using RAPAd® CMV Adenoviral Expression System (Cell Biolabs, Inc., San Diego, CA). Generation of the full-length human KGA and GAC constructs was described in our prior publication [12]. Adenoviral constructs were amplified in a 293 AD cell line (Cell Biolabs) and purified by ultracentrifugation through a CsCl gradient. Viral titer was determined by the Adeno-X™ Rapid Titer Kit (Clontech Laboratories, Inc., Mountain View, CA).

EVs were isolated from the supernatants of GLS1-overexpressing cells, HIV-1-infected macrophages, and LPS-activated microglia through differential centrifugations. Briefly, the supernatants were first centrifuged at 300×g for 10 min to remove free cells, at 3000×g for 20 min to remove cellular debris, and then 10,000×g for 30 min to remove free organelles. Lastly, EVs were collected by ultracentrifugation at 100,000×g for 2 h at 4 °C. To prepare EVs for Western blotting, the EV pellets were lysed in M-PER mammalian protein extraction reagent (Thermo Scientific, Pittsburgh, PA). For negative staining, EVs were fixed in 2% glutaraldehyde and 2% paraformaldehyde. For glutaminase activity assay and neurotoxicity, the EVs were resuspended in 1 ml of glutamine-free neurobasal medium.

EV isolations from the brains were carried out as described previously with modifications according to the protocol [22]. The fresh and previously frozen mice hemibrains were harvested and dissected finely. The brain samples were then treated with 20 units/ml papain (Worthington) in Hibernate E solution (BrainBits, Springfield, IL) for 15 min at 37 °C. The same volume of cold Hibernate E solution was added to the brain samples to stop the reaction of papain. The brain tissue was then gently homogenized and filtered through a 40-μm mesh filter (BD Biosciences), followed by a centrifugation at 300×g for 10 min and 3000×g for 20 min at 4 °C to get rid of cells, membranes, and debris. After the supernatants were filtered through 0.45-μm filter (Thermo Scientific), they were subjected to 10, 000×g for 30 min at 4 °C to eliminate organelle contaminations. The supernatants were further centrifuged at 100,000×g for 70 min at 4 °C to pellet EVs. The pellets were then resuspended in filtered PBS, or MPER lysate solution for NanoSight or Western blot. All the samples were ultracentrifuged in ultraclear polycarbonate tubes (Beckman Coulter) that have a volume of 13.2 ml. A Beckman Coulter ultracentrifuge (Beckman Coulter OptimaL-90K ultracentrifuge; Beckman Coulter, Fullerton, CA, USA) was used with a rotor type SW 41 Ti.

EVs were fixed and then spread on the silicon monoxide and nitro-cellular film-coated copper grid. The droplets were removed with filter paper, air-dried at room temperature, and then subjected to transmission electron microscopy (TEM).

A NanoSight NS 300 (Malvern) equipped with an sCMOS camera was utilized to analyze the size distribution and concentration of EVs. NanoSight utilizes NTA, which is a combination of light scattering and Brownian motion technology to measure the concentration and size and distribution of particles in the EV supernatants. After the whole process of EV isolation, the pellets were first resuspended in 100 μl of filtered PBS and then diluted 100 times. The conditions of the measurements include temperature of 25 °C; viscosity of 1 cP, 25 s per capture frame; and a measurement time of 60 s. All the conditions were kept the same among all the samples. The results indicate the mean sizes and concentration of at least three individual measurements.

Protein concentrations were determined by Bradford protein assay. SDS PAGE separated proteins from the whole cell and EV lysates. Afterward, they were electrophoretically transferred to polyvinyldifluoridene membranes (Millipore, Billerica, MA and Bio-Rad, Hercules, CA). The membranes were incubated overnight at 4 °C with polyclonal antibodies for KGA and GAC (Dr. N. Curthoys, Colorado State University, Fort Collins, CO), tissue transglutaminase (tTG) (Lab Vision/Thermo, Fremont, CA), flotillin-2 (Cell Signaling Technology, Danvers, MA), and β-actin (Sigma), followed by horseradish peroxidase-linked secondary anti-rabbit or anti-mouse secondary antibodies (Cell Signaling Technology). Antigen-antibody complexes were visualized by Pierce ECL Western Blotting Substrate. For quantification of the data, films were scanned with a CanonScan 9950F scanner, and images were analyzed using the public domain NIH Image program (developed at the US National Institutes of Health).

Highly concentrated whole cell lysates were collected from flasks and subjected to GLS activity assay using a two-step assay [27, 47]. Briefly, protein concentrations in the lysates were tested by using BCA Protein Assay Kit (Pierce). All samples were normalized to same concentration. In the first step, 50 mg of protein were added to 100 μl of initial assay mix. The mix contains 50 mM glutamine, 0.15 M phosphate, 0.2 mM EDTA, and 50 mM Tris-acetate. The PH value of the mix was adjusted to 8.6 and incubated at 37 °C for 30 min. Ten microliters of 3 N hydrochloric acid was added to inactivate the glutaminase activity and stop the reaction. In the second step, 1 ml of the second reaction mix was added, which contained 0.4 mg of purified bovine liver glutamate dehydrogenase (Sigma-Aldrich, St. Louis, MO, USA), 0.08 M Tris-acetate at pH 9.4, 0.2 M hydrazine, 0.25 mM adenosine 5′-diphosphate sodium salt, and 2 mM β-nicotinamide adenine dinucleotide hydrate. The samples were mixed and incubated for 30 min at room temperature. One hundred microliters of the reaction was used for measurement, absorbance was determined at a wavelength of 340 nm, and glutamate concentration was determined using a standard curve of 10, 5, 2.5, 1.25, 0.625, and 0.0 mM glutamate, along with negative controls.

Glutamate levels were analyzed by RP-HPLC using an Agilent 1200 liquid chromatograph and fluorescence detector as previously described [14] with a few modifications. The experiments utilized 4.6 × 75 mm, 3.5 μm ZORBAX Eclipse AAA analytical columns (Agilent). A gradient elution program was optimized for glutamate measurement with a flow rate 0.75 ml/min. The intracellular glutamate levels in the whole brain lysates of mice and whole cell lysates were determined by Amplex Red Glutamic Acid/Glutamate Oxidase Assay Kit (Invitrogen) based on the manufacturer’s instruction. The brain tissue lysates and whole cell lysates were diluted to the same protein concentration before the assay.