P2X7R influences tau aggregate burden in human tauopathies and shows distinct signalling in microglia and astrocytes

Abstract

The purinoceptor P2X₇R is a promising therapeutic target for tauopathies, including Alzheimer's disease (AD). Pharmacological inhibition or genetic knockdown of P2X₇R ameliorates cognitive deficits and reduces pathological tau burden in mice that model aspects of tauopathy, including mice expressing mutant human frontotemporal dementia (FTD)-causing forms of tau. However, disagreements remain over which glial cell types express P2X₇R and therefore the mechanism of action is unresolved. Here, we show that P2X₇R protein levels increase in human AD post-mortem brain, in agreement with an upregulation of P2RX7 mRNA observed in transcriptome profiles from the AMP-AD consortium. P2X₇R protein increases mirror advancing Braak stage and coincide with synapse loss. Using RNAScope we detect P2RX7 mRNA in microglia and astrocytes in human AD brain, including in the vicinity of senile plaques. In cultured microglia, P2X₇R activation modulates the NLRP3 inflammasome pathway by promoting the formation of active complexes and release of IL-1β. In astrocytes, P2X₇R activates NFκB signalling and increases production of the cytokines CCL2, CXCL1 and IL-6 together with the acute phase protein Lcn2. To further explore the role of P2X₇R in a disease-relevant context, we expressed wild-type or FTD-causing mutant forms of tau in mouse organotypic brain slice cultures. Inhibition of P2X₇R reduces insoluble tau levels without altering soluble tau phosphorylation or synaptic localisation, suggesting a non-cell autonomous role of glial P2X₇R on pathological tau aggregation. These findings support further investigations into the cell-type specific effects of P2X₇R-targeting therapies in tauopathies.

Keywords: Alzheimer’s disease; Astrocyte; Human brain; Microglia; P2X(7)R; RNAScope; Tauopathy.

Fig. 1

P2X₇R expression is elevated prior to end-stage AD. Representative immunoblots using an antibody against P2X₇R of total brain homogenates from (a) BA9 prefrontal cortex and (b) BA21 temporal cortex. β-actin was included as a loading control. Bar charts show quantification of P2X₇R protein relative to β-actin in each sample as percentage of average control (Braak 0-II). Representative immunoblots of the post-synaptic and pre-synaptic markers (c,e) PSD-95 and (d, f) synaptophysin (SYP) in BA9 and BA21 total homogenates. Bar charts show quantification of synaptic proteins relative to neuron-specific enolase (NSE) in each sample as percentage of average control (Braak 0-II). Data is mean ± SEM. Following (a,c,d) d’Agostino and Pearson or (b,e,f) Shapiro-Wilk normality tests, data was analysed using a parametric one-way ANOVA with Dunnett́s multiple comparison test or a non-parametric Kruskal-Wallis test with Dunńs multiple comparison test. (a,c,d) n = 25 (Braak 0-II), 19 (Braak III-IV), 16 (Braak V-VI). (b,e,f) n = 4 (Braak 0-II), n = 5 (Braak III-IV), n = 5 (Braak V-VI). *p < 0.05. g,h) Volcano plots showing changes in expression of P2R transcripts in temporal cortex from (g) AD and (h) PSP brain versus age-matched controls obtained from the Mayo RNA-seq cohort of the AMP-AD consortium. Dotted lines indicate the cut-off p-value of 0.05.

Fig. 2

P2RX7 mRNA localises to GFAP⁺ astrocytes and CD68⁺ microglia surrounding Aβ plaques in AD brain. a) Representative in-situ hybridisation-immunohistochemistry (ISH-IHC) images of brain sections from frontal cortex (BA9) of human control brain (Braak stage 0-II) hybridised with RNAscope probes against P2RX7 mRNA (red puncta, indicated by arrowheads) and immunolabelled with antibodies against GFAP and βIII-tubulin (blue/grey chromogenic stain) and CD68 (diaminobenzidine [DAB] stain). n = 3. Representative images of P2RX7 RNAscope performed in sections from three Braak stage III-IV and three Braak stage V-VI AD cases immunolabelled with antibodies against (b) GFAP, (c) CD68 and (d) Aβ (6E10). Arrowheads indicate representative P2RX7 mRNA puncta in GFAP or CD68 immunopositive cells, or in the vicinity of Aβ deposits. Insets indicate regions of each image displayed at a higher magnification. n = 3. Scale bar: 10 µm. e) Volcano plots showing changes in expression of P2r transcripts in transgenic CRND8 mouse brains versus non-transgenic controls at 3, 6, 12 and 20 months obtained from the AMP-AD Knowledge Portal (https://doi.org/10.7303/syn3157182). Dotted lines indicate the cut-off p-value of 0.05. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Microglial P2X₇R activation induces the formation of ASC specks and release of IL-1β. a) Schematic depicting the stimulation of microglia with 100 ng/mL LPS for 3 h, after which medium was replaced and cells pre-treated with 0.1% DMSO (vehicle) or 1 µM Cp2 for 1 h prior to the addition of 1 mM ATP for 20 min. Untreated and LPS-primed cells were included as additional controls. b) Representative confocal images of microglia immunolabelled using an antibody against ASC (green) under basal conditions (untr.), following stimulation with LPS only, LPS and ATP only, or with LPS and ATP in the presence of vehicle (DMSO) or Cp2. ASC specks are indicated by white arrowheads. Hoechst-33342 was used as to stain nuclei. Insets indicate representative regions of each image displayed at higher magnification. Scale bar: 50 µm. c) Quantification of the number of ASC specks normalised to the number of Hoechst⁺ nuclei per condition, displayed as a percentage relative to vehicle-treated cells exposed to LPS + ATP (n = 4). d) Bar graph shows quantification of the amounts of IL-1β in the supernatant of microglial cultures primed with LPS, followed by pre-treatment with vehicle or Cp2 and stimulated with ATP. Untreated, LPS-primed and LPS + ATP only conditions were also included (n = 3). e) Representative cytokine array of cytosolic fractions isolated from post-mortem BA9 AD and control brain at different Braak stages (0-II, III-IV, V-VI). IL-1β coordinates are indicated in purple. f) Quantification of IL-1β amounts in Braak stage III-IV and V-VI BA9 displayed relative to Braak 0-II tissues. n = 10 per group (Braak 0-II, III-IV, V-VI). Following Shapiro-Wilk normality test, data was analysed using (c,d) one-way ANOVA with Dunnett’s multiple comparison test or (f) Kruskal-Wallis test with Dunńs multiple comparison test. Data is mean ± SEM. *p < 0.05, **p < 0.01,****p < 0.0001. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

Astrocytic P2X₇R regulates the expression of Lcn2, activates NFκB and induces cytokine production. a) mRNA levels of Lcn2 in astrocytes stimulated with BzATP (300 µM, 4 h) expressed as relative abundance to control (n = 4). b) Representative immunoblots of Lcn2 and Aldh1L1 in control or BzATP stimulated astrocytes. Bar graph shows the quantification of Lcn2 levels normalised to Aldh1L1 in BzATP stimulated astrocytes expressed as percentage of control (n = 6). c) mRNA levels of Lcn2 in astrocytes pre-treated with 0.1 % DMSO (vehicle) or the P2X₇R antagonist Cp2 (1 µM) for 1 h prior to stimulation with BzATP expressed as relative abundance to control (n = 4). d) Representative immunoblots of Lcn2 and Aldh1L1 in vehicle-treated or astrocytes treated with the indicated concentrations of Cp2 for 1 h prior to stimulation with BzATP. Bar graph shows the quantification of Lcn2 normalised to Aldh1L1 expressed relative to vehicle (n = 3–4). e) Representative images of NFκB labelling (green) in control and BzATP stimulated astrocytes (upper panel). Hoechst-33342 was used as a nuclear stain. Merged images are displayed in the lower panel. Scale bar: 50 µm. Bar chart shows quantification of the mean nuclear intensity of NFκB relative to mean total cell intensity per well expressed relative to control (n = 3). f-g) Representative immunoblots of Ser536 phosphorylated (p-)NFκB p65 subunit and total NFκB (p65) in (f) BzATP-stimulated and control astrocytes and (g) unstimulated astrocytes, treated with vehicle or with the indicated concentrations of Cp2 for 1 h prior to stimulation with BzATP. Bar graph displays the quantification of p-NFκB normalised to NFκB relative to (f) untreated (n = 6) or (g) vehicle (n = 3–4). h-j) mRNA levels of (h)ccl2 (n = 5),(i)cxcl1 (n = 4) (j)il-6 (n = 5) in control astrocytes and astrocytes stimulated with BzATP, treated with vehicle or Cp2 (1 µM) for 1 h prior to stimulation with BzATP. Data is expressed as relative abundance to vehicle. k) Representative cytokine array from BA9 control and AD brain at different Braak stages (0-II, III-IV, V-VI). CCL2 and IL-6 coordinates are indicated in purple. n = 10 per group. l-m) Quantification of (l) CCL2 and (m) IL-6 amounts in Braak III-IV and V-VI AD BA9 relative to Braak 0-II. Following a Shapiro-Wilk normality tests, (a,c,h-j) p-values were obtained from log-transformed values using unequal variance Welch́s t-test. Data is mean ± SD. Data analysed using (b,e,f) unpaired t-test, (d,g) one-way ANOVA with Dunnett́s multiple comparison test or (l, m) Kruskal-Wallis test with Dunńs multiple comparison test. Data is mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5

P2X₇R inhibition reduces tau aggregate levels in BSCs expressing P301L/S320F human tau. a) Volcano plots showing changes in expression of P2r transcripts in cortex from rTg4510 mice versus non-transgenic controls at 2.5, 4.5 and 6 months, obtained from the AD-AMP Knowledge Portal (https://doi.org/10.7303/syn3157182). Dotted lines indicate the cut-off p-value of 0.05. b) Schematic represents the preparation of organotypic brain slice cultures (BSCs) from CD1 mice at P7-P8. BSCs were transduced with 1x10¹¹ vg/mL rAAV2/8 expressing EGFP-tagged 0N4R WT or (P301L/S320F) human tau (hTau) under the hybrid cytomegalovirus enhancer chicken β-actin (hCBA) promoter at 0 DIV. Region-matched slices were treated with 10 µM Cp2 at 14 DIV and every 2–3 days thereafter with each media change until 28 DIV. Control slices were treated identically with vehicle (0.1 % DMSO). Representative confocal images of BSCs transduced with EGFP-tagged WT or P301L/S320F-hTau at 28 DIV. Scale bar: 50 µm. Representative immunoblots of low-speed supernatant (LSS), high-speed supernatant (HSS) and sarkosyl-insoluble (SI) fractions probed with antibodies against (c) total tau (DAKO), β-actin and (d) tau phosphorylated at Ser396/404 (PHF1). Bar graphs display the quantification of (c) insoluble tau, determined as the proportion of total tau in the SI fraction relative to total tau in LSS from the same sample and (d) phosphorylated insoluble tau, calculated as the amount of PHF1 relative to total tau in the SI fraction, shown relative to vehicle-treated slices expressing WT-hTau. Representative immunoblots of synaptoneurosome (SNS) and cytosolic (cyt) fractions of BSCs immunoblotted with antibodies against (e) total tau, PSD-95 and (f) PHF1. Bar charts display the ratio of (e) total tau and (f) PHF1-immunoreactive tau normalised to total tau in SNS relative to the cyt compartment, expressed as percentage relative to BSCs expressing WT-hTau and treated with DMSO. Following Shapiro-Wilk normality test, data was analysed using two-way ANOVAs with Sidaḱs multiple comparison test. n = 3. Data is mean ± SEM. *p < 0.05, **p < 0.01,***p < 0.001.