TNF-α promotes extracellular vesicle release in mouse astrocytes through glutaminase

C57BL/6J mice were housed and maintained in the Comparative Medicine Facility of the Tongji University School of Medicine (Shanghai, China). All procedures were conducted in accordance with the protocols approved by the Institutional Animal Care and Use Committee at the Tongji University School of Medicine. Postnatal (P1 to P2) brain tissues were used for mouse astrocyte cultures. Gls knock in mouse embryonic stem (ES) cells were obtained from the Knockout Mouse Project (KOMP) Repository (CA, USA). Recombinant mouse TNF-α and IL-1β were obtained from R&D Systems. N-Acetyl-l-cysteine (NAC, A7250), 6-Diazo-5-oxo-l-norleucine (L-DON, D2141), BPTES (SML0601), and GW4869 (D1692) were obtained from Sigma-Aldrich.

Cortices of C57BL/6J mice were dissected and mechanically dissociated using forceps to remove the membranes and large blood vessels. Brain tissues were digested by Trypsin-EDTA (Life Technologies) and then plated on cell culture flasks in Dulbecco’s modified Eagle medium Nutrient Mixture F-12 (DMEM/F-12). Culture medium was supplemented with FBS (10% v/v) and penicillin/streptomycin (1% v/v). Cultures were maintained in a humidified chamber (37 °C, 5% CO₂ incubator). After 7 to 10 days, the astrocytes were harvested by trypsinization.

The method for the isolation of extracellular vesicles has been described previously [24]. EVs were isolated from the serum-free culture of mouse astrocytes through differential centrifugation. 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 intracellular organelles. Lastly, EVs were collected by ultracentrifugation at 100,000 × g for 2 h at 4 °C. To prepare EVs for western blot, the EVs pellets were lysed in M-PER mammalian protein extraction reagent (Thermo Scientific, Pittsburgh, PA).

Extracellular vesicles were characterized at 25 °C using Nano ZS90 (Malvern Instruments, UK). Eighty microliter samples were loaded into a microcuvette (ZEN0118, Malvern Instruments, UK) for measurement.

The size and number of extracellular vesicles were assessed with NanoSight NS300 system (Malvern Instruments, UK). Astrocytes were cultured in 6-cm culture dishes. At 24 h after medium change, EVs were isolated from normalized volumes of serum-free culture supernatants through differential centrifugation and resuspended with 150 μl PBS. The supernatant was diluted at 1:100 in PBS, and 1 ml solution was used for NanoSight analysis.

Mouse astrocytes were grown on a glass coverslip, fixed with 2.5% glutaraldehyde, and washed three times with PBS. The cells were then dehydrated in a series of increasing ethanol concentrations and transferred for critical drying. After coating with platinum/palladium using a sputter coater, the sample was imaged with a scanning electron microscope (S-3400, Hitachi).

EVs were negatively stained and then spread on the copper grids. The droplets of EVs were removed with filter paper and air-dried at room temperature. Images were obtained using transmission electron microscopy (JEM-1230, JEOL Ltd.).

Cells or EV pellets were lysed in M-PER mammalian protein extraction reagent (Thermo Scientific). 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 treated with purified primary antibodies for glutaminase (1:1000; Abcam), GFAP (1:1000; Cell Signaling Technologies), β-actin (1:5000; Sigma Aldrich), flotillin-2 (1:5000; BD biosciences), and ALG-2 interacting protein (Alix) (1:1000; Cell Signaling Technologies) overnight at 4 °C followed by a horseradish peroxidase-linked secondary anti-rabbit or anti-mouse antibody (1: 5000; Icllab). Antigen–antibody complexes were visualized by Pierce ECL Western Blotting Substrate. Mitochondrial protein was isolated by the Mitochondria Isolation Kit for Cultured Cells (ab110171, Abcam). ATP5A (1:1000; Abcam) was used as a mitochondrial marker to indicate that there is no mitochondria contamination in the cytosol.

Mouse astrocytes cultured on cover glasses were fixed with 4% PFA, rinsed with PBS, and then blocked by 2% BSA in PBS. Cells were incubated overnight at 4 °C with primary antibodies anti-GFAP (1:1000; Abcam). Cover glasses were washed and incubated for 1 h at room temperature with secondary antibodies including anti-mouse IgG (coupled with Alexa Fluor 488, Life Technologies). Nuclear DNA was stained with DAPI. Cover glasses were mounted on glass slides with mounting buffer (Sigma-Aldrich). Morphological changes were visualized by a Zeiss 710 confocal laser scanning microscope.

ROS measurement was assayed by dichloro-dihydro-fluorescein diacetate (DCFH-DA). Mouse astrocytes were incubated in 10 μM DCFH-DA (50101, YEASEN) for 30 min at 37 °C, 5% CO₂ and then were washed with PBS. The ROS were determined using conventional fluorescence microscopy (ZEISS).

Intracellular glutamate detection was performed with the Amplex Red Glutamic Acid/Glutamate Oxidase Assay Kit from Invitrogen following manufacturer’s procedure. High performance liquid chromatography (HPLC) analysis for extracellular glutamate was performed as previously described [25].

Briefly, mouse astrocytes were treated with TNF-α for 24 h, the cell viability was assayed by cell viability assay kit (Promega, G7570). The luminescent signal recording from the reaction was as the standard of cell viability.

Data were evaluated statistically by the analysis of variance (ANOVA), followed by Tukey’s test for multiple comparisons. Data were shown as mean ± SD. *, **, and *** denote p < 0.05, p < 0.01, and p < 0.001 in comparison to control, respectively. ^#, ^##, and ^### denote p < 0.05, p < 0.01, and p < 0.001 in comparison to TNF-α-treated groups, respectively.

Mouse astrocytes were established from the cortices of neonatal mice as previously described (Additional file 1: Figure S1a) [26]. To visualize EV generation in astrocytes, we first used scanning electron microscopy (SEM) on the cultures. SEM revealed polarized membranous structures of mouse astrocytes that are likely EVs in the course of releasing from the plasma membrane (Fig. 1a).

Our previous studies have demonstrated HIV-1 infection and immune activation increase release of EVs from macrophages and microglia [24]. To further assess the relationship between proinflammatory cytokines and the release of extracellular vesicles, we used dynamic light scattering (DLS) analysis, which determined the size distribution of small particles in suspension. In particular, the release of EVs from astrocytes during inflammatory cytokine treatment was examined. By analyzing the signaling intensity (Additional file 1: Figure S1d), we found that EV sizes ranged from 50 to 800 nm, and TNF-α (50 ng/ml) treatment had no effect on the size of EVs. Cell viability did not change after TNF-α treatment (Additional file 1: Figure S1b). To determine quantitative changes of EVs, we isolated them from serum-free supernatants with or without TNF-α treatment for 24 h. The EV pellets were collected and resuspended in PBS for negative staining under TEM (Fig. 1b, c). The numbers of EVs per field under TEM were significantly higher in TNF-α treatment group as compared with that in control (Fig. 1d), suggesting that TNF-α treatment leads to increased release of EVs in mouse astrocytes. Calreticulin, a marker of endoplasmic reticulum, is not present in the 100,000 × g fraction, suggesting that cell debris is not present in the fraction (Additional file 1: Figure S1c). EVs were also subjected to Western Blot for specific EVs markers, including ALG-2 interacting protein (Alix), HSP70, Flotillin-1, TSG101, and Flotillin-2 (Fig. 1e). Consistent with TEM data, TNF-α significantly increased the expression levels of Alix (Fig. 1f), HSP70 (Fig. 1g), Flotillin-1 (Fig. 1h), TSG101 (Fig. 1i), and Flotillin-2 (Fig. 1j) in EVs from the treated astrocytes as compared to those from the control. In addition, we also analyzed EVs using NTA tracking. NTA tracking allowed a precise analysis of secreted vesicle size and concentration (Fig. 1k). Consistent with the previous data, TNF-α induced a two-fold increase in the concentrations of EVs, as compared to the control group (Fig. 1m), but the average size of the EVs had no change (Fig. 1l). Together, these data suggest that TNF-α increases EV release in primary mouse astrocytes.

Previous studies showed that TNF-α regulates the expression of GLS in human neuron and microglia [5, 27]. Therefore, we further investigated whether glutaminase, particularly its two main isoforms KGA and GAC, is upregulated by TNF-α treatment in mouse astrocytes. Protein levels of GLS isoforms in astrocyte cultures were determined following proinflammatory cytokine treatment (TNF-α 50 ng/ml, IL-1β 10 ng/ml). When protein levels of GLS isoforms were determined by Western Blot, TNF-α induced a 1.8-fold increase of GAC (58 kDa) when compared with the untreated control (Fig. 2a, b), whereas IL-1 β had no significant effect on protein levels of GAC (Fig. 2a, b). Therefore, we used TNF-α to treat mouse astrocytes in the subsequent experiments. To determine whether changes of GLS lead to changes of enzyme activity, we checked the intracellular and extracellular glutamate levels in mouse astrocytes upon TNF-α treatment. As expected, TNF-α dramatically upregulated intracellular glutamate as compared to the untreated control group (Fig. 2c). Similar to the intracellular glutamate level, the extracellular glutamate level also increased significantly following the treatment with TNF-α as compared to the control (Fig. 2d). Interestingly, we found that the location of GLS in mouse astrocytes had changed after TNF-α treatment and GLS were released from the mitochondria to the cytoplasm (Additional file 2: Figure S2a). Together, these data suggest that TNF-α increases GLS expression that leads to an increase of its enzymatic production of glutamate in mouse astrocyte.

Mitochondria are a major source of intracellular ROS, and are also particularly vulnerable to oxidative stress. Our previous studies suggest that ROS is closely related to mitochondrial permeability [28, 29]. To determine the relationship between oxidative stress and GLS, mouse astrocytes were pretreated with 1 mM L-DON, a glutaminase inhibitor. After 6 h of treatment with or without TNF-α, the ROS level in the cells was detected by 2,7-dichlorodi-hydrofluoresceindiacetate (DCFH-DA), a common indicator for ROS generation in cells [30, 31]. The intracellular ROS was determined by the conversion of non-fluorescence DCFH-DA to a fluorimetric DCF that exhibits peak fluorescence in the green spectrum. TNF-α treatment increased ROS generation in astrocytes as compared with untreated control group (Fig. 3a, b). Furthermore, inhibition of GLS by L-DON, a competitive glutaminase inhibitor, blocked the ROS generation (Fig. 3a). We also used an allosteric GLS inhibitor BPTES, one of the most potent blockers of GLS, to inhibit GLS activity. BPTES suppressed the production of ROS, which increased in TNF-α-treated astrocytes (Fig. 3b, c). In addition, when we pretreated the astrocytes with NAC, a remover of ROS, the ROS levels in TNF-α-treated astrocytes were reversed (Fig. 3b, c). These data suggested that ROS generation is correlated with GLS activity in mouse astrocytes.

To determine whether GLS participated in the release of EVs in TNF-α-treated mouse astrocytes, we pretreated the cultures with BPTES or GW4869 (a neutral sphingomyelinases inhibitor that halts the biogenesis of EVs for 1 h. After TNF-α treatment, EVs were extracted and then subjected to Western Blot for EV detection (Fig. 4a). After the densitometric quantifications of the EV markers TSG101 (Fig. 4b), Alix (Fig. 4c), and Flotillin-2 (Fig. 4d), we found that they were significantly increased in EVs isolated from TNF-α-treated astrocytes. Both BPTES and GW4869 dramatically inhibited the EV release, and the expression levels of the EV markers decreased significantly when compared with TNF-α group (Fig. 4b–d). Together, these interesting findings indicated that GLS was an important factor participating in TNF-α-mediated EV release.

Previous research has shown that the release of EVs was closely related to ROS [32‐34]. In order to further investigate the relationship between ROS and EV release in mouse astrocytes, cells were pretreated with NAC (10 mM) for 1 h, and then EVs were collected from serum-free culture after 24 h treatment. Pretreatment with NAC dramatically reduced the levels of TSG101 (Fig. 4b), Alix (Fig. 4c), and Flotillin-2 (Fig. 4d) in the EV lysates, indicating that the release of EVs in TNF-α group was blocked by NAC (Fig. 4).

Taken together, these data suggested that GLS and the subsequent ROS generation was an important pathway for EV release in mouse astrocytes during TNF-α-mediated inflammation.

To investigate GLS in mediating EV release in mouse astrocytes, Gls knock out (Gls−/−) mice were used. Since Gls−/− mice died shortly after birth [35], Gls−/− astrocytes were obtained at E18.5. Protein samples of Gls−/− astrocytes were used for Western Blot (Fig. 5a) to confirm gene knockout. The ROS production in the GLS knockout mice was detected (Additional file 2: Figure S2b), and the quantification data showed the ROS production had no change with TNF-α stimulation (Additional file 2: Figure S2c). Wild-type and Gls−/− mouse astrocytes were first treated with TNF-α for 24 h, and EVs were extracted for Western Blot (Fig. 5b). After the densitometric quantifications of the Western Blot, the data revealed significantly increased levels of Alix (Fig. 5c) and Flotillin-2 (Fig. 5d) in EV lysates from TNF-α treated wild type cells. Furthermore, after TNF-α treatment, the Gls−/− group had reduced levels of EVs as compared to wild-type cells (Fig. 5c, d). EVs derived from wild-type and Gls−/− mouse astrocytes were also subjected to size-distribution analysis by NTA (Fig. 5e). TNF-α treatment induced a two-fold increase of EVs in the wild type astrocytes. Knockout of GLS induced a ten-fold decrease in EV concentration compared with wild-type mouse astrocytes (Fig. 5e). These findings demonstrated that GLS is important for the release of extracellular vesicles in mouse astrocytes following TNF-α treatment.

Taken together, our data unveiled a relationship between glutaminase and EV release in mouse astrocytes. TNF-α treatment upregulated protein levels of GLS and increased the production of ROS. GLS and ROS could influence each other, and both of them could promote EV release.