The Neuroprotective Effect of a Specific P2X₇ Receptor Antagonist Derives from its Ability to Inhibit Assembly of the NLRP3 Inflammasome in Glial Cells

Abstract

Release of interleukin (IL)‐1β from immunocompetent cells requires formation of the NACHT, LLR and PYD domains‐containing protein 3 (NLRP3) inflammasome and caspase 1 activation. Adenosine 5′‐triphosphate (ATP), acting on the P2X₇ receptor, is one factor that stimulates inflammasome assembly. We show that a novel specific P2X₇ receptor antagonist, GSK1370319A, inhibits ATP‐induced increase in IL‐1β release and caspase 1 activation in lipopolysaccharide (LPS)‐primed mixed glia by blocking assembly of the inflammasome in a pannexin 1‐dependent manner. GSK1370319A also inhibits ATP‐induced subregion‐specific neuronal loss in hippocampal organotypic slice cultures, which is dependent on its ability to prevent inflammasome assembly in glia. Significantly, GSK1370319A attenuates age‐related deficits in long‐term potentiation (LTP) and inhibits the accompanying age‐related caspase 1 activity. We conclude that inhibiting P2X₇ receptor‐activated NLRP3 inflammasome formation and the consequent IL‐1β release from glia preserve neuronal viability and synaptic activity.

INTRODUCTION

Microglia are the immunocompetent cells of the central nervous system (CNS) and react profoundly to changes in the microenvironment. In their so‐called resting state, microglial processes are motile and constantly make contact with neighboring cells; it is estimated that they sample the entire parenchyma every few hours (17). Because of the multiple receptors expressed on microglia, they respond to numerous stimuli 2, 10, 30 and upon stimulation, adopt one of several activated phenotypes. In general, it seems that strong or prolonged stimulation leads to release of inflammatory cytokines and chemokines whereas controlled stimulation, for example, in response to acute injury, may lead to release of neurotrophins and exert a restorative effect (25).

Under physiological conditions, intracellular concentration of the purine, adenosine 5′‐triphosphate (ATP), is between 3 and 10 mM, and extracellular concentration is maintained between 400 and 700 nM (23). However, under stress conditions, for example, following ischemia, ATP is released from damaged cells and extracellular concentration markedly increases, allowing ATP to act as a danger‐associated molecular pattern (DAMP). The P2X₇ receptor is a member of the purinergic receptor family, a group of receptors that are activated by ATP (26). It is reported that millimolar concentrations of extracellular ATP, which occur only during injury or inflammation, are required to activate P2X₇ receptors (7). These receptors have been identified specifically on cells of the immune system including macrophages (24), mast cells (4) and are also expressed on microglia (31). Stimulation of the receptor initiates transmembrane ion fluxes and formation of a large, reversible membrane pore that allows the passage of molecules of up to 900 kDa (26); this pore has been identified as pannexin 1 (21).

The P2X₇ receptor is involved in the processing and release of the proinflammatory cytokine interleukin‐1β (IL‐1β), and the evidence suggests that this involves the formation of the NACHT, LLR and PYD domains‐containing protein 3 (NLRP3) inflammasome 7, 15. This is a collection of cytosolic proteins including apoptotic speck protein (ASC), the Nod‐like protein NLRP3 and caspase 1. The formation of the inflammasome leads to processing of caspase 1 to its active form that subsequently cleaves pro‐IL‐1β to its mature form enabling its release from the cell (13). Since IL‐1β appears to be a potential mediator of neurodegeneration, for example, in ischemia (29), epilepsy (19) and Alzheimer's disease (6), the role of the inflammasome in neurodegenerative disorders is a subject of significant interest.

Here, we report that a specific P2X₇ antagonist, GSK1370319A, inhibits the ATP‐induced release of IL‐1β from mixed glia, but not IL‐6 or tumor necrosis factor α (TNFα), and the action is a consequence of preventing the assembly of the NLRP3 inflammasome. ATP markedly decreases hippocampal cell viability in an organotypic slice culture, and this is mimicked by IL‐1β and blocked by the P2X₇ antagonist. Culturing neurons with ATP‐treated glia decreases cell viability and that this is attenuated when glia are treated with ATP and the P2X₇ antagonist. These data indicate that the release of IL‐1β from glia, which is dependent on assembly of the inflammasome, markedly affects cell viability. Significantly, we have identified a compound that inhibits formation of the inflammasome and which, as a consequence, preserves neuronal viability.

MATERIALS AND METHODS

GSK1370319A

GSK1370319A is a novel and recently identified P2X₇ selective antagonist; the structure is presented in Figure 1. Using a fluorescent imaging plate reader (FLIPR) assay, the pIC50 for the P2X₇ receptor was established as >7.5, with selectivity over P2X_2/3, P2X₁ and P2X₄ receptors (pIC50 ≤ 4.5 in all cases), as well as a range of other receptors including 5HT, cannabinoid, cholinergic and adrenergic receptors.

Figure 1.

Structure of the novel P2X7 receptor antagonist, GSK1370319A.

Primary mixed glial cultures

Mixed glia were prepared from the cortices of 1‐day‐old Wistar rats (Trinity College, Dublin, Ireland). Cortical tissue was cross‐chopped, incubated for 25 minutes at 37°C in Dulbecco's modified Eagle's medium (DMEM, Invitrogen, Paisley, UK) supplemented with 10% fetal bovine serum (Invitrogen) and 50 U/mL penicillin/streptomycin (Invitrogen) and plated (2.5 × 10⁵) as previously described (18). Cells from each rat were treated separately. After 13 days, cells were primed with lipopolysaccharide (LPS; 1 µg/mL; Sigma, Dublin, Ireland). After a 4‐h period of incubation in the presence of LPS, cells were treated with ATP (Sigma) and incubation continued for a further 24 h. In some experiments, LPS‐treated cells were incubated in the presence GSK1370319A (1 µM–20 µM GlaxoSmithKline, Harlow, UK) for 1 h prior to addition of ATP. In one series of experiments, to knock down the panx1 gene, mixed glia were incubated in the presence or absence of panx1 siRNA (Dharmacon, Lafayette, CO, USA) and DharmaFECT 1 transfection reagent (Dharmacon) for 48 h. Western blot analysis was used to confirm knockdown of pannexin 1 protein. In all experiments, supernatants were collected, and cells were harvested for later analysis.

To prepare purified microglia and astrocytes, cells were grown in T25 flasks in DMEM as above. After 12 days, the flasks were shaken for 2 h at 110 rpm, at room temperature and tapped several times to remove nonadherent microglia. The supernatants were removed from the flask and centrifuged at 2000 rpm for 3 min at 21°C. The pellet was resuspended in DMEM and the cells were counted. Cells were pipetted into six‐well plates at a density of 1 × 10⁵ cells/mL. To prepare astrocytes, the flasks containing the adherent astrocytes were washed with phosphate‐buffered saline (PBS) and 1 mL of 0.05% w/v trypsin‐ethylene diamine tetraacetic acid (EDTA) was added at 37°C until the cells just began to detach, DMEM was then added to the flask to inhibit the action of trypsin. The cells were centrifuged at 2000 rpm for 3 min. The pellet was resuspended in DMEM and the cells were plated in nine‐well plates at a density of 1 × 10⁵ cells/mL. Isolated microglia and astrocytes were treated in the same manner as mixed glial cultures.

Primary neuronal cultures

Neurons were prepared from cortices of 1‐day‐old Wistar rats. Tissue was cross‐chopped and incubated in Hank's balanced salt solution (Invitrogen), buffered with N‐(2‐hydroxyethyl)piperazine‐N′‐(2‐ethanesulfonic acid) (HEPES) and dissociated using papain (both from Sigma). Cells were plated at 6 × 10⁴ cells/mL on a 24‐well plate precoated with poly‐D‐lysine and maintained in neurobasal medium (Invitrogen) containing 2 mM glutamine (Sigma) and 2% B27 without antioxidants (Invitrogen, Ireland). Neurons were cocultured with primary mixed glial cells using Greiner Thinserts^® (Cruinn Diagnostics, Dublin, Ireland). Cell viability was assessed using a MTS assay (Promega, Southampton, UK) on the supernatants of the neuronal cultures.

Organotypic hippocampal slice cultures

Organotypic hippocampal cultures were prepared and maintained as previously described (9). Briefly, 4‐ to 5‐day‐old Wistar rats were decapitated, the brain removed and placed in dissection media containing Hank's balanced salt solution (Gibco, Carlsbad, CA, USA), 20 mM HEPES and 6.5 mg/mL d‐glucose (Sigma). The hippocampi were dissected out, cut into 400‐µm slices using a McIllwain tissue chopper (Mickle, Laboratory, Guildford, UK) and transferred into fresh dissection media. Slices for culture were selected based on the presence of intact hippocampal subfields (CA1, CA3 and the dentate gyrus). The slices were transferred onto porous (0.4 µm) membranes of a millicell insert® (Millipore, Cork, Ireland), placed into 24‐well tissue culture plates containing culture medium (300 µL; 50% minimal essential media (Invitrogen), 25% horse serum, 2% B27, 4 mM glutamine, 6 mg/mL d‐glucose and 50 U/mL penicillin/streptomycin) and incubated in a humidified chamber with 5% CO₂ at 37°C for 10 days. Slices were incubated in the presence of LPS (1 µg/mL) and after 4 h, GSK1370319A (20 µM) or the caspase 1 inhibitor Z‐WEHD‐FMK (300 µM) was added; incubation continued for 2 h after which time ATP (5 mM) was added and incubation continued for a further 24 h. Cell death was assessed by propidium iodide (PI) uptake, which enters cells with damaged membranes, binds to nucleic acids and becomes brightly fluorescent. PI (5 µg/mL) was added to the cultures for 40 min, which were washed with fresh media and examined using a Hammamatsu Orca 285 camera attached to an Olympus IX51 under EX/EM 540 to 580/600 to 660 nm (red). PI staining was quantified in each subregion of the hippocampus by analysis of optical density using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Mean values for each subregion of the hippocampus were obtained from five field views (1 µm²) and data from six similarly treated slices are presented.

Analysis of IL‐1β, IL‐6 and TNFα concentrations

Supernatant concentrations of IL‐1β, IL‐6 and TNFα obtained from glial, neuronal and organotypic slice cultures were measured using enzyme‐linked immunosorbent assay (ELISA; Duoset, R&D Systems Europe, Abingdon, UK or BD OptiEIA, BD Biosciences, Oxford, UK). Cytokine concentrations in the test samples were evaluated with reference to the standard curves prepared using recombinant cytokines of a known concentration.

Analysis of IL‐1β, IL‐6 and TNFα mRNA

RNA was isolated from glia using Nucleospin^® RNAII KIT (Macherey‐Nagel, Duren, Germany) and cDNA was prepared using High‐Capacity cDNA RT kit according to the manufacturer's instructions (Applied Biosystems, Paisley, UK). Real‐time polymerase chain reaction (PCR) for the detection of IL‐1β (Mm00443258_m1), TNFα (Mm00443258_m1) and IL‐6 (Mm00446191_m1) mRNA was performed with predesigned Taqman gene expression assays (Applied Biosystems). mRNA was normalized to the endogenous control, actin RNA. Samples were assayed on an Applied Biosystems 7500 Fast Real‐Time PCR machine.

Analysis of proteins by Western immunoblotting

Western blotting was performed as previously described (12). Cultured cells were harvested, homogenized in buffer containing Tris‐HCl (0.01 M) and EDTA (1 mM), and protein (20 µg) was boiled in gel‐loading buffer and separated by 12% sodium dodecyl sulfate‐polyacrylamide gel electrophoresis. Proteins were transferred to nitrocellulose membranes and incubated with antibodies against the following: β‐actin (mouse monoclonal; Sigma), pannexin 1, caspase 1 p10, pan 14–3‐3, Bax (all rabbit polyclonal, Santa Cruz Biotechnology, Santa Cruz, CA, USA), phospho c‐Jun N‐terminal kinase (pJNK), cleaved caspase 3, cleaved caspase 7 and cleaved caspase 8 (Cell Signalling Technology, Beverly, MA, USA). Membranes were incubated with horseradish peroxides‐conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) and bands were visualised using Supersignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL, USA). Images were captured using a Fujifilm LAS‐3000 (Brennan and Co, Dublin, Ireland). Densitometry was performed using ImageJ software (http://rsbweb.nih.gov/ij/).

Immunoprecipitation

For immunoprecipitation studies, mixed glial samples containing 500 µg protein were incubated overnight at 4°C in the presence of ASC specific antibody (5 µg; rabbit polyclonal; Santa Cruz Biotechnology) or IgG isotype control (negative control) (Sigma Aldrich, Dublin, Ireland). A/G protein agarose beads (50 µL; Santa Cruz Biotechnology) were added, samples were incubated for 2 h, washed in PBS containing NP‐40 (0.01%) and centrifuged. Loading buffer (1X) was added to each sample, and samples were boiled. Proteins were separated (see above) and NLRP3 was visualized by incubating the membrane in the presence of a NLRP3‐specific antibody (rabbit polyclonal; Santa Cruz Biotechnology).

Analysis of activity of caspases 1 and 3

Cells were harvested, washed PBS and lysed in lysis buffer. Activities of caspases 1 and 3 were measured using fluorometric assay kits (R&D Systems Europe Ltd., Abingdon, UK).

Chronic treatment of young and aged animals with GSK1370319A

Groups of young (3 months, n = 6 per treatment) and aged (22 months; n = 8 per treatment) male Wistar rats (Harlan, Blackthorn, UK) were randomly assigned to control‐treated and GSK1370319A‐treated groups. All rats were housed in groups of three and maintained under veterinary supervision in a controlled environment (12‐h light schedule; ambient temperature 22–23°C) and all experiments were performed under a license issued by the Department of Health (Ireland) and in accordance with the guidelines laid down by the local ethical committee. Animals had free access to food (standard laboratory chow; Red Mills, Kilkenny, Ireland) and water and body weights were recorded throughout the study. GSK319 (10 mg/kg twice daily) was administered orally in maple syrup (3 mg/0.2 mL) for 56 days, while rats in the control groups received only maple syrup.

Induction of long‐term potentiation (LTP) in vivo

On day 56 of treatment, 24 h after the last dose of GSK1370319A, rats were anesthetized by intraperitoneal injection of urethane (1.5 g/kg). When necessary, a top‐up dose of urethane, to a maximum of 2.5 g/kg, was given to achieve deep anesthesia; this was indicated by the absence of a pedal reflex. The ability of rats to sustain LTP in perforant path‐granule cell synapses, in response to tetanic stimulation of the perforant path was assessed as previously described (16).

At the end of the experiment, rats were killed by cervical dislocation and the brains were rapidly removed and snap frozen for later analysis.

Statistical analysis

Data were analyzed as appropriate using either Student's t‐test for independent means or analysis of variance (ANOVA) followed by post hoc Tukey's test to determine which conditions were significantly different from each other. Data are expressed as means ± standard error of the mean (SEM).

RESULTS

ATP induces activation of caspase 1 and release of the pro‐inflammatory cytokine IL‐1β, but not IL‐6 or TNFα, from LPS‐primed mixed glia

Incubation of mixed glia in the presence of LPS significantly increased IL‐1β mRNA (**P < 0.01; ANOVA; Figure 2A); this was not modulated by the presence of ATP at any concentration. In contrast, mixed glial cells that were primed with LPS for 6 h and treated with varying concentrations of ATP released IL‐1β in a dose‐dependent manner; maximum release was observed following incubation with 5 mM ATP (*P < 0.05; ***P < 0.001; ANOVA; Figure 2B). The addition of 5 mM ATP to unprimed mixed glia exerted no significant effect. ATP also dose‐dependently increased caspase 1 activation, in LPS‐primed cells as assessed by a fluorometric assay; the highest concentration, 5 mM ATP, exerted a marked increase (***P < 0.001; ANOVA; Figure 2C), although 2 mM ATP also significantly increased enzyme activity (*P < 0.05; ANOVA).

Figure 2.

Adenosine 5′‐triphosphate (ATP; 5 mM) significantly increased interleukin (IL)‐1β release (B) and caspase 1 activation (C) [***P < 0.001; analysis of variance (ANOVA)], but not IL‐1β mRNA (A) in lipopolysaccharide (LPS)‐primed mixed glia. Pretreatment with GSK1370319A (GSK319) (1 and 20 µM) dose‐dependently inhibited ATP‐induced changes in IL‐1β release (E) and caspase 1 activation (F) (⁺ P < 0.05; ⁺⁺ P < 0.01; ⁺⁺⁺ P < 0.001; ANOVA; LPS + ATP + GSK1370319A vs. LPS + ATP), but not IL‐1β mRNA (D). Panx1 siRNA reduced expression of pannexin 1 compared with scrambled siRNA (G) and significantly decreased the ATP‐induced IL‐1β release (H; ***P < 0.001; ANOVA) in LPS‐primed cells in a manner similar to GSK1370319A (H; ⁺⁺⁺ P < 0.001; ANOVA; LPS + ATP + Panx1 siRNA or GSK1370319A vs. LPS + ATP). ATP‐induced binding of ASC to NACHT, LLR and PYD domains‐containing protein 3 (NLRP3) was decreased by GSK1370319A (I).

The release and transcription of two further proinflammatory cytokines, IL‐6 and TNFα was investigated. LPS increased both mRNA and supernatant concentration of both cytokines (**P < 0.01; ***P < 0.001; ANOVA; data not shown); ATP exerted no modulatory effect on these LPS‐induced changes.

GSK1370319A, a novel specific P2X₇ antagonist, inhibits the ATP‐induced caspase 1 activation and release but not transcription of IL‐1β

LPS‐primed mixed glia were treated in the presence or absence of a novel P2X₇ antagonist, GSK1370319A, prior to treatment with ATP and, consistent with the data described in Figure 2A,B, IL‐1β mRNA and IL‐1β release from LPS‐primed cells were significantly increased by ATP (*P < 0.001; ANOVA; Figure 2D,E). Incubation of cells in the presence of GSK1370319A inhibited the ATP‐induced effect on IL‐1β release, with a return to near control concentration with the higher concentration (⁺⁺ P < 0.01; ⁺⁺⁺ P < 0.001; ANOVA; Figure 2E). However, GSK1370319A did not attenuate the ATP‐induced increase in IL‐1β mRNA in LPS‐primed cells (Figure 2D). Similarly the significant increases in release of IL‐6 and TNFα induced by LPS + ATP (*P < 0.05; ***P < 0.001; ANOVA) were unaffected by GSK1370319A (data not shown). Because the effect of GSK1370319A was specific to IL‐1β release, we considered that it may affect inflammasome activity, and therefore activation of caspase 1 was assessed. Incubation of cells in the presence of GSK1370319A dose‐dependently inhibited the effect of ATP (5 mM) with the highest concentration exerting a complete block (⁺ P < 0.05; ⁺⁺⁺ P < 0.001; ANOVA; Figure 2F). Similar changes in IL‐1β release (LPS + BzATP − 1544.73 ± 188.27 pg/mL; LPS + BzATP + GSK1370319A − 567.83 ± 104.08 pg/mL) and caspase 1 activation (LPS + BzATP − 8.42 ± 0.83 a.u.; LPS + BzATP + GSK31370319A − 3.44 ± 0.61 a.u.) were also seen when cultures were treated with the specific P2X₇ agonist BzATP (100 mM) instead of ATP (data not shown).

Pannexin 1 is essential for the ATP‐induced release of IL‐1β and GSK1370319A inhibits assembly of the inflammasome

Pannexin 1 is the protein that forms the transmembrane pore that opens following stimulation of the P2X₇ receptor allowing the passage of molecules of molecular weight up to 900 kDa (21). In order to assess whether pannexin 1 was necessary for ATP‐induced release of IL‐1β, the panx1 gene was knocked down using an siRNA directed toward the gene. Treatment of cells with siRNA reduced the expression of pannexin 1 by approximately 60% compared with control cells that were treated with a scrambled siRNA (***P < 0.001; Student's t‐test for independent means; Figure 2G). Mixed glial cells in which panx1 had been knocked down were primed with LPS and treated with ATP but in these cells ATP was unable to induce significant release of IL‐1β, contrasting with the significant effect observed in control‐treated cells (⁺⁺⁺ P < 0.001; LPS + ATP vs. LPS + ATP + Panx1 siRNA; ANOVA; Figure 2H). The modulatory effect of panx1 siRNA was similar to that of GSK1370319A, which also attenuated the LPS + ATP‐induced IL‐1β release (⁺⁺⁺ P < 0.001; ANOVA; Figure 2H).

ASC is one of the component proteins of the NLRP3 inflammasome assembly which results in caspase 1 cleavage and subsequent cleavage of IL‐1β to its mature, active form. Accordingly, we immunoprecipitated ASC from LPS‐primed mixed glia that were treated with ATP alone, or in combination with GSK1370319A, and assessed binding of NLRP3. While ATP induced substantial binding of ASC to NLRP3 (Figure 2I), NLRP3 precipitation with ASC was reduced by about 75% in cells treated with ATP and GSK1370319A (***P < 0.001; Student's t‐test for independent means; Figure 2I).

The P2X₇ receptor is expressed on and mediates its effects predominantly through microglial cells

In order to identify the cell on which the P2X₇ antagonist exerted its effects, we prepared isolated microglial and astrocytic cultures. Western blot analysis indicated that the P2X₇ receptor is predominantly expressed in microglia, although a low level of expression was observed in astrocytes (Figure 3A). The data show that ATP triggered a significant increase in release of IL‐1β, and a significant increase in activation of caspase 1, in LPS‐primed astrocytes (**P < 0.01; ***P < 0.001; ANOVA; Figure 3B,C) and that these changes were partially attenuated by 20 µM GSK1370319A (⁺ P < 0.05; ANOVA), but unaffected by 1 µM GSK1370319A. ATP also induced IL‐1β release and caspase 1 activation in LPS‐primed microglia (***P < 0.001; ANOVA; Figure 3B,C) and these changes were partially attenuated by 1 µM and completely attenuated by 20 µM GSK1370319A (⁺⁺ P < 0.01; ⁺⁺⁺ P < 0.001; ANOVA; Figure 3B,C). The stimulatory effect of ATP was substantially greater in LPS‐primed microglia compared with astrocytes, in line with the expression profile of P2X₇R.

Figure 3.

Western blot analysis indicated that the expression of P2X₇R was greater in isolated microglial compared with astrocytic cultures (A). Adenosine 5′‐triphospha (ATP, 5 mM) significantly increased interleukin (IL)‐1β release (B) and caspase 1 activation (C) in lipopolysaccharide (LPS)‐primed microglial and astrocytic cultures [**P < 0.01; ***P < 0.001; analysis of variance (ANOVA); control vs. ATP]. Pretreatment with GSK1370319A (GSK319) (1 and 20 µM) dose‐dependently inhibited these ATP‐induced changes (⁺ P < 0.05; ⁺⁺ P < 0.01; ⁺⁺⁺ P < 0.001; ANOVA).

ATP and recombinant IL‐1β induce cell death and apoptotic caspase activation in hippocampal organotypic slice cultures

Having established that ATP‐induced inflammasome formation in LPS‐treated mixed glia, we evaluated its effect in LPS‐primed hippocampal organotypic slice cultures. ATP induced significant cell death in the dentate gyrus and CA3 pyramidal regions of the hippocampal slice culture as measured by uptake of PI (**P < 0.01; ***P < 0.001; ANOVA; Figure 4A,B,C); ATP‐induced cell death in CA1 was substantially less but was still significantly greater than control (*P < 0.05; ANOVA; Figure 4D). To assess whether cell death might be because of IL‐1β released from glia, LPS‐primed organotypic slice cultures were treated with 5 ng/mL of recombinant IL‐1β, and this mimicked the effect of 5 mM ATP in dentate gyrus and CA3 (**P < 0.01; ***P < 0.001; ANOVA; Figure 4A,B,C), but did not affect viability of CA1 pyramidal cells (Figure 4D). IL‐1β‐induced cytotoxicity was dose dependent (data not shown). Hippocampi were microdissected into dentate gyrus, CA3 and CA1 regions and IL‐1R1 expression in each region was assessed. It was found that IL‐1R1 was expressed predominantly in the dentate gyrus and CA3 regions of the hippocampus but minimally in the CA1 region (Figure 4E).

Figure 4.

Adenosine 5′‐triphosphate (ATP) and interleukin (IL)‐1β significantly increased propidium iodide (PI) uptake into cells of the dentate gyrus (A,B), CA3 (A,C) and CA1 (A,D) in LPS‐primed organotypic slices cultures [*P < 0.05; **P < 0.01; ***P < 0.001; analysis of variance (ANOVA)], paralleling IL‐1R1 expression (E). ATP and IL‐1β induced cleavage of effector caspase‐3 and ‐7 but not caspase‐8 (F).

Western blot analysis was used to assess activation of caspases in the hippocampal slice cultures after treatment with IL‐1β and ATP. Both induced cleavage of effector caspases 3 and 7, but there was no effect on cleavage of the death receptor‐associated caspase 8 (Figure 4F).

ATP induced cell death in organotypic slice cultures is attenuated by treatment with the P2X₇ antagonist GSK1370319A and the caspase 1 inhibitor Z‐WEHD‐FMK

To assess whether activation of the P2X₇ receptor was involved in ATP‐induced cytoxicity, hippocampal slice cultures were treated with ATP and GSK1370319A. The findings confirmed that ATP significantly increased PI uptake in dentate gyrus, CA3 and, to a lesser extent, CA1 (**P < 0.01; ***P < 0.001; ANOVA; Figure 5A–D) and revealed that GSK1370319A significantly reduced the cell death induced by ATP to nearly control levels (⁺ P < 0.05; ⁺⁺⁺ P < 0.001; ANOVA; Figure 5A–D). As P2X₇ activation results in activation of caspase 1, we investigated whether inhibition of caspase 1 was also capable of attenuating ATP‐induced cell death (⁺⁺ P < 0.01; ⁺⁺⁺ P < 0.001; ANOVA). The data show that the specific caspase 1 inhibitor Z‐WEHD‐FMK also significantly attenuated the ATP‐induced cytotoxicity (⁺⁺⁺ P < 0.001; ANOVA; Figure 5) to the same extent as the P2X₇ antagonist.

Figure 5.

Adenosine 5′‐triphosphate (ATP)‐induced propidium iodide (PI) uptake [A–D; **P < 0.01; ***P < 0.001; analysis of variance (ANOVA)] was significantly attenuated by GSK1370319A (GSK319; 20 µM) and the specific caspase 1 inhibitor Z‐WEHD‐FMK (300 µM) (A–D; ⁺⁺ P < 0.01; ⁺⁺⁺ P < 0.001; ANOVA). ATP increased phosphorylation of JNK and 14‐3‐3, reduced binding of the proapoptotic protein Bax to 14‐3‐3, and increased expression of cleaved caspase 1, ‐7 and ‐3; these changes were attenuated by GSK319 and Z‐WEHD‐FMK (E).

We next investigated the activation of caspases 1, 3 and 7 in slice cultures treated with ATP and GSK1370319A or Z‐WEHD‐FMK. As previously seen in mixed glial cultures, treatment with ATP increased activation of caspase 1, and this was attenuated by GSK1370319A and the caspase 1 inhibitor Z‐WEHD‐FMK. ATP‐induced activation of caspases 3 and 7 was also attenuated by GSK1370319A, and by Z‐WEHD‐FMK (Figure 5E). ATP triggered phosphorylation of JNK and 14‐3‐3, and it reduced binding of the proapoptotic protein Bax to 14‐3‐3.

Soluble factors released from mixed glia affect neuronal viability

To investigate whether the P2X₇ inhibitor reduced neuronal cytotoxicity by inhibiting the release of IL‐1β from glial cells, cocultures of glia and neurons were set up in which each cell type was grown in close proximity to each other but were separated by a 0.4‐µm membrane. In this experiment, only the mixed glial cells were treated but neurons were exposed to soluble factors released from glia. First, we confirmed that treatment of mixed glia with ATP significantly released IL‐1β and increased caspase 1 activity (***P < 0.001; ANOVA) and that these changes were attenuated by GSK1370319A and Z‐WEHD‐FMK (⁺⁺⁺ P < 0.001; ANOVA; Figure 6A,B). Using a trans‐well system, neurons were cultured with mixed glial from these four treatment groups. When neurons were cocultured with ATP‐treated cells, the concentration of IL‐1β in the media in which neurons were bathed was significantly increased compared with control (***P < 0.001; ANOVA; Figure 6C), but there was no evidence of a similar increase in the media when neurons were cultured with mixed glia that were treated with ATP + GSK1370319A or ATP + Z‐WEHD‐FMK (⁺⁺⁺ P < 0.001; ATP alone vs. ATP + GSK1370319A or ATP + Z‐WEHD‐FMK; ANOVA; Figure 6C). These changes were closely paralleled by changes in neuronal caspase 3 activation. Thus neurons that were grown under ATP‐treated mixed glia exhibited a significantly greater caspase 3 activation than those grown under control‐treated glia (***P < 0.001; ANOVA; Figure 6D) and there was no evidence of a similar change in activation of caspase 3 in the neurons cultured with glia that were treated with ATP + GSK1370319A or ATP + Z‐WEHD‐FMK (⁺⁺⁺ P < 0.01; ATP alone vs. ATP + GSK1370319A or ATP + Z‐WEHD‐FMK; ANOVA; Figure 6D). In addition to caspase 3 activation, neurons that were cultured with ATP had reduced viability as measured by MTS assay (***P < 0.001; ANOVA; Figure 6E) and cell viability was restored when neurons were cultured with mixed glia that were treated with ATP and GSK1370319A or ATP and Z‐WEHD‐FMK (⁺⁺⁺ P < 0.001; ANOVA). Direct treatment of neurons with ATP had no effect on caspase 3 activation or on their viability as assessed by the MTS assay (data not shown).

Figure 6.

Lipopolysaccharide (LPS)‐primed mixed glia were treated with adenosine 5′‐triphosphate (ATP) in the presence/absence of GSK1370319A (GSK319) or Z‐WEHD‐FMK. ATP significantly increased interleukin (IL)‐1β release (A) and caspase 1 activity (B) [***P < 0.001; analysis of variance (ANOVA)] and these effects were inhibited by 20 µM GSK319 and 300 µM Z‐WEHD‐FMK (⁺⁺⁺ P < 0.001; ATP vs. ATP + GSK319 or ATP + Z‐WEHD‐FMK). Neurons were cocultured in Greiner Thinserts^® (Cruinn Diagnostics, Dublin, Ireland) (which prevents physical interaction of the cells) with LPS‐primed glia which were treated with ATP alone or in combination with GSK319 or Z‐WEHD‐FMK. IL‐1β concentration in the neuronal media (C) and neuronal caspase‐3 activation (D) were significantly increased and neuronal viability was significantly decreased (E) when ATP‐treated glia were present in the coculture (***P < 0.001; ANOVA), but these changes were absent when neurons were cocultured with mixed glia that were treated with ATP + GSK319 or ATP + Z‐WEHD‐FMK (⁺⁺⁺ P < 0.001; ATP alone vs. ATP + GSK319 or ATP + Z‐WEHD‐FMK; ANOVA).

Chronic treatment with GSK1370319A attenuates age‐related deficits in LTP

We considered that the neuroprotective effect of GSK1370319A observed in organotypic slices might extend to an in vivo model and therefore evaluated its effect on the ability of aged rats to sustain LTP, as the age‐related impairment has been linked with a deficit in neuronal function (11). Here we demonstrate that LTP was significantly decreased in control‐treated aged rats compared with young rats (***P < 0.001 young vs. aged; ANOVA; Figure 7A,B). Treatment of aged rats with GSK1370319A for 56 days improved their ability to sustain LTP so that the mean percentage change in field excitatory postsynaptic potential (EPSP) slope in the last 10 minutes of the experiment (compared with the value in the 10 minutes before tetanic stimulation) was 99.81% ± 0.96% and 110.58% ± 0.96% in aged control‐ and GSK1370319A‐treated rats, respectively (⁺⁺⁺ P < 0.001 aged control vs. aged GSK1370319A treated; ANOVA; Figure 7A,B), compared with 118.62% ± 2.53% in young rats. There was no significant difference between young control and young GSK1370319A‐treated rats (P > 0.05; ANOVA; Figure 7B). In order to assess inflammasome formation we examined activation of caspase 1 in hippocampal tissue prepared from these rats; there was a significant increase in caspase 1 activation in aged compared with young rats (**P < 0.01; ANOVA; Figure 7C) which was attenuated in tissue prepared from aged GSK1370319‐treated rats (⁺ P < 0.05; ANOVA).

Figure 7.

Long‐term potentiation (LTP) in dentate gyrus was reduced in aged compared with young control‐treated rats [A,B; ***P < 0.001; analysis of variance (ANOVA)] but this was attenuated in GSK1370319A (GSK319)‐treated aged rats (⁺⁺⁺ P < 0.001; ANOVA aged vs. aged GSK319 treated). Caspase 1 activity was increased in tissue prepared from aged, compared with young, animals [C: **P < 0.01; ANOVA] and this was attenuated by chronic treatment with GSK319 (⁺ P < 0.05; ANOVA).

DISCUSSION

The specific P2X₇ inhibitor GSK1370319A blocks NLRP3‐dependent recruitment of ASC and therefore assembly of the inflammasome in glia. As a result, it attenuates the effect of ATP on caspase 1 activation and IL‐1β release in LPS‐primed mixed glia. Release of IL‐1β from glia exerts a marked effect on viability of hippocampal neurons in an organotypic slice culture. This neuronal loss is dependent on the dissociation of Bax from phosphorylated 14‐3‐3 resulting in activation of the effector caspases 3 and 7. Inhibiting glial P2X₇ receptors prevents the sequence of events leading to neurodegeneration.

LPS increased mRNA expression and release of IL‐1β, IL‐6 and TNFα in mixed glial cells and treatment of LPS‐primed cells with ATP further increased release of IL‐1β, confirming earlier reports in bone marrow‐derived macrophages 22, 27 and mixed glial cultures (1). We report that the effect was IL‐1β‐specific, with no ATP‐induced change in IL‐1β mRNA or in expression or release of IL‐6 or TNFα, and while pretreatment of cells with GSK1370319A inhibited the increase in IL‐1β, it did not affect IL‐1β mRNA or release of IL‐6 or TNFα. It has been reported previously that ATP can attenuate the effect of LPS on release of IL‐6 and TNFα in microglia under different experimental circumstances; thus, coincubation with ATP (1 mM) decreased LPS (100 ng/mL)‐induced release of these cytokines (3), whereas we observed that ATP (5 mM) was unable to modulate the effect of LPS (1 µg/mL) when cells were preincubated in its presence for 6 h.

Given that the modulatory effect of GSK1370319A described here was confined to IL‐1β, and because it has been reported that P2X₇ activation induces formation of the NLRP3 inflammasome in bone marrow‐derived macrophages 22, 27, we examined caspase 1 activity. In parallel with the changes in IL‐1β release, ATP increased caspase 1 activity in LPS‐primed mixed glia, and this was inhibited by GSK1370319A. This concurs with finding that ATP failed to induce IL‐1β release and caspase 1 activity in LPS‐primed macrophages prepared from P2X₇ ^−/− mice (20). Importantly, we found that the effect of ATP on LPS‐primed release of IL‐1β and on caspase 1 activity in mixed glia, as well as the modulatory effect of GSK1370319A, were also evident in isolated microglia and, to a much lesser extent, in isolated astrocytes. The differential effects in isolated cells reflect the markedly greater expression of P2X₇ receptors on microglia.

The importance of pannexin in IL‐1β release in glia is demonstrated by the finding that knockdown of the panx1 gene suppressed the ATP‐induced P2X₇‐induced release of IL‐1β. This indicates P2X₇ activation leads to formation of the pannexin‐containing pore in glia, supporting the evidence of a similar finding in macrophages 21, 26. Importantly, the present results also demonstrate that release of IL‐1β is dependent on an interaction between NLRP3 and ASC and that GSK1370319A prevents this interaction and inhibits assembly of the inflammasome.

To evaluate the effect of increased glial‐derived IL‐1β on neurons, we prepared organotypic hippocampal slice cultures and first investigated the effect of direct application of ATP. It markedly decreased viability of cells especially in the dentate gyrus and CA3, to a lesser extent, in CA1. Application of IL‐1β exerted the same effect with less marked changes in viability of CA1 neurons compared with neurons in the dentate gyrus and CA3. To date, no studies have focused on the vulnerability of specific hippocampal neuronal populations to inflammatory cytokines, although the findings parallel the IL‐1R1 expression, which is relatively greater in the dentate gyrus and CA3, compared with CA1; in situ hybridization studies revealed similar relative expression (5). The ATP‐induced changes described here appear to be a consequence of activation of effector caspase 7 and 3 triggered by phosphorylation of the chaperone protein 14‐3‐3 and its release of the proapoptotic protein Bax. The importance of P2X₇ receptor activation was highlighted by the finding that GSK1370319A blocked the ATP‐induced cytotoxicity and the accompanying activation of caspase 7 and 3. The finding that inhibition of caspase 1 also blocked the action of ATP on cell viability confirms a pivotal role for IL‐1β.

Coculturing neurons with LPS‐primed glia that were treated with ATP with or without GSK1370319A or caspase 1 inhibitor (in a manner that prevented physical interaction of the cells), established that soluble factors released from ATP‐treated glia profoundly affected neuronal viability and that this was coupled with activation of caspase 3. The effect was IL‐1β mediated because inhibiting caspase 1 blocked the ATP‐induced effect on cell viability and therefore adds to other evidence that indicates a neurotoxic effect of IL‐1β(28). Preventing the activation of P2X₇ also blocked the detrimental effects of ATP‐treated glia on neurons and the findings are consistent with the proposal that it does so by inhibiting IL‐1β release. This highlights a neuroprotective effect that is conferred by inhibition of P2X₇ receptors. This neuroprotective effect may be an important factor in the partial restoration of LTP in aged GSK1370319A‐treated animals, in which the age‐related increase in caspase 1 activation was attenuated.

The importance of the inflammasome in the pathogenesis of diseases ranging from rheumatoid arthritis to amyotrophic lateral sclerosis has been highlighted in the past few years 8, 14 and, clearly, an understanding of the mechanisms that control its assembly may have far‐reaching consequences for identifying the next generation of therapeutic strategies. Our evidence indicates that inflammasome formation and the subsequent IL‐1β release from glia are triggered by ATP‐induced activation of P2X₇ receptors. The data reveal that release of IL‐1β following assembly of the inflammasome has a marked effect on viability of hippocampal neurons and that this is linked with release of the proapoptotic protein Bax from its chaperone protein 14‐3‐3 and subsequent downstream activation of caspase 7 and 3. Of particular significance is the demonstration that inhibiting P2X₇ activation by GSK1370319A blocks the detrimental effect of ATP‐activated glia on neuronal viability and partially restores synaptic plasticity in aged animals. Therefore, we have identified a neuroprotective effect of this compound GSK1370319A that is secondary to its ability to inhibit inflammasome assembly in glia. We propose that P2X₇ receptor antagonists may play a major role in modulating the progression of diseases of the CNS in which neuroinflammatory changes are identified.