Macrophage expression of P2X7 controls autoimmune uveitis

Abstract

The release of danger-associated molecular patterns (DAMPs) such as ATP from stressed or damaged cells is a key initiator of sterile inflammation. In autoimmune diseases, extracellular ATP acts as a potent pro-inflammatory signal by activating the purinergic receptor P2X7, which is expressed on both mononuclear phagocytes (MPs) and T cells—key drivers of pathological processes. While it is well established that P2X7 mediates ATP-dependent immune activation, its cell-specific contributions to innate versus adaptive immunity in autoimmune conditions remain unclear, especially in autoimmune uveitis. Here we used the experimental autoimmune uveoretinitis (EAU) mouse model to delineate the cellular mechanisms underlying P2X7’s role in autoimmune responses. Using a combination of multiple cell-specific conditional models and transcriptomic approaches, we showed a pivotal role forP2X7 expressed by MPs in orchestrating T-cell mediated autoimmune responses. P2X7 deficiency in MPs decreased disease severity. Additionally, cell-specific transcriptomic analyses, including single-cell analyses, revealed that P2X7 exerted distinct modulatory effects across monocyte-derived macrophages (MdM) versus microglia. In MdM, lack of P2X7 was associated with reduced expression of genes related to the inflammasome, phagocytosis pathways, and components of the complement system, leading to a marked decrease in pathogenic Th17 cell frequency in the retina. In microglia, P2X7 deficiency instead particularly impacted an IFN-responsive microglial subset that is normally characteristic of EAU. By specifically deleting P2X7 in microglia, we demonstrate its role in driving pathogenic processes in this cell population. These findings suggest that inhibition of P2X7 could be a promising therapeutic strategy in autoimmune neuroinflammatory disorders.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12974-025-03529-w.

Highlights

1. P2X7 expressed on MPs, rather than on CD4⁺ T-cells, plays a primary role in the development of EAU.

2. Lack of P2X7 expression by MPs not only down-regulated phagocytosis and inflammasome pathways but also complement components.

3. P2X7 indirectly influences Th17 polarization through its expression in MPs.

4. P2X7-deficiency in MPs reduced the frequency of the IFN-responsive microglial subset during EAU.

5. P2X7 expression by microglia contributes to the development of EAU.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12974-025-03529-w.

Introduction

Danger-associated molecular patterns (DAMPs) are released by damaged or dying cells to activate the innate immune system. While this sterile inflammation promotes tissue repair, it can also contribute to the development of inflammatory diseases including cancer, autoimmune diseases and neurodegenerative diseases [1]. Among the immune receptors that recognize DAMPs, the P2X7 receptor is activated by high concentration of ATP released from the intracellular content of cells (mM range) in contrast to the extracellular space, where its normal concentration is much lower (nM range).

P2X7 is a potent activator of the NLRP3 inflammasome, leading to the cleavage of pro-IL-1β and pro-IL-18 and the release of the mature forms by mononuclear phagocytes (MPs) [2, 3]. Moreover, P2X7 mediates cleavage of cell surface proteins such as TNFα by ADAM17 [4, 5] and the production of reactive oxygen species (ROS) and IL-6 [6–10]. P2X7 is also an important regulator of T cell functions and was shown to be involved in T-cell activation, proliferation, and cell death depending on T-cell subsets, P2X7 expression level, and the amount of ATP [10–12]. Several data also provide strong evidence for the participation of P2X7 in Th1/Th17 polarization, but the direct or indirect role of P2X7 via neighboring cells is still undetermined [2]. In addition, our previous studies have shown that P2X7 is also involved in the production of the chemokines by glial cells suggesting an indirect role of P2X7 in T-cell recruitment [13, 14]. Hence, the role of P2X7 has been studied extensively in the context of infection and inflammation. The evidence suggests that P2X7 enhances the immune system’s response against pathogen infections, triggers microbicidal mechanisms, and helps controlling infection. Conversely, in inflammatory and neurodegenerative diseases P2X7 contributes to pathological processes [2, 10, 15].

Nevertheless, the function of P2X7 in the context of autoimmune diseases involving innate and adaptive immune cells remains poorly understood. P2X7 expression is frequently elevated in autoimmune diseases, and its inhibition has generally been associated with beneficial outcomes [15]. These effects were linked to a reduction of the NLRP3 inflammasome activation and/or a decreased proportion of Th17 cells [16, 17]. In experimental autoimmune encephalomyelitis (EAE), a specific P2X7 tracer proved to be a valuable marker of neuroinflammation and microglial activation, underscoring the receptor’s potential contribution to disease pathogenesis [18]. However, the specific cellular target(s) and precise mechanisms through which P2X7 contributes to disease development require further investigation.

Systemic or organ-specific autoimmune diseases are caused by dysregulation of the immune system leading to the activation of autoreactive lymphocytes. The mechanisms by which autoreactive T cells trigger pathology have been characterized in various autoimmune models. The disease is first initiated by activation of CD4⁺ T-cells in secondary lymphoid organs and their differentiation into T helper 1 (Th1) and Th17 cells. In the second phase, these autoreactive T-cells infiltrate the target organ and mediate the recruitment and activation of MPs. Activated MPs amplify the inflammatory response through cytokine production, antigen presentation, and tissue-destructive mechanisms, playing a central role in mediating tissue damage [19]. Finally, the disease progression is modulated by the suppressive functions of regulatory T-cells (Tregs). However, the identification of therapeutic targets to control this T-cell mediated inflammation is still the subject of intense research.

Among autoimmune diseases, non-infectious uveitis (NIU) is a group of inflammatory intraocular diseases which can be debilitating and lead to blindness [20]. NIU can be related to systemic auto-inflammatory disorders, such as Sarcoidosis, Behçet’s disease, Vogt-Koyanagi-Harada disease, or can be eye-limited [21]. Some patients are refractory to the usual corticosteroid-based treatments or have severe side-effects with long-term treatment, stressing the need for new therapeutic targets [22]. The pathogenesis of NIU involves dysregulated cytokine networks, with elevated levels of IL-1β, IL-6, and TNF-α detected in serum and/or aqueous humor samples [23, 24]. While anti-TNF-α antibodies are currently used in clinical management, emerging therapies targeting other inflammatory mediators are under active investigation to achieve more effective immunomodulation [25]. Interestingly, in Behçet’s disease patients, monocytes exhibit elevated P2X7 expression and activation, resulting in elevated IL-1β secretion [26]. These findings suggest that dysregulated P2X7 signaling may exacerbate autoimmune-driven pathology in this context. Thus, P2X7 could represent a promising therapeutic target; however, a deeper understanding of its functions across both innate and adaptive immunity within the ocular environment is required before clinical translation. In the present study, we explored the cell-specific role of P2X7 in the pathogenesis of autoimmune neuroinflammation using experimental autoimmune uveoretinitis (EAU), a model of autoimmune uveitis [19]. Using cell specific knock-out mice, we showed that disruption of P2X7 in MPs in general led to a notable reduction of the clinical severity of the disease and infiltration of leucocytes, while lack of P2X7 expression in T-cells or in Tregs specifically did not alter the development of the disease. Furthermore, we found that P2X7 deletion differentially affected MdMs and microglia. In MdMs, P2X7 deficiency reduced the expression of genes related to pro-inflammatory mediators and the phagocytosis pathway. In contrast, P2X7 loss impaired a pathogenic IFN-responsive microglial subset, and targeted deletion of P2X7 in microglia was sufficient to decrease EAU severity, highlighting its pathogenic role in EAU. In summary, our study revealed a major role of P2X7 expressed by MPs during autoimmune neuroinflammation.

Materials and method

Animals

Both female and male mice aged 8 to 12 weeks were used for this project. Wild-type C57BL/6J mice were purchased from Janvier Laboratories. P2X7 knock-out (P2X7^−/−) mice generated by Gabel’s group at Pfizer were obtained from the Jackson Laboratory (strain # 005576) [27]. P2X7^lox/lox P2X7^lox) mice were engineered as previously described [28]. P2X7^lox mice were crossed with Foxp3^creER mice (strain # 016961) [29], Cx3cr1^cre mice (strain #025524), or Cx3cr1^creER mice (strain #020940) [30], all obtained from the Jackson Laboratory, to generate cell-specific conditional P2X7 knockout and P2X7^lox control mice. All cre-expressing mice were heterozygous at the cre locus. All mice were negative for the Crb1^rd8 mutation, to avoid confusion of retinal phenotypes due to this mutation. The animals were housed in cages in groups of 5 with enrichment on ventilated racks (temperature and hygrometry automatically regulated) with a 12 h/12 h lighting cycle. After the arrival of the mice, a week of acclimatization was respected before experimentation.

Treatment with tamoxifen

Tamoxifen (Sigma) was dissolved in corn oil (Sigma) at a concentration of 20 mg/ml. Mice received daily intraperitoneal injections of tamoxifen (100 mg/kg) for 5 consecutive days.

Induction of EAU

For active induction of EAU, mice were immunized subcutaneously with 200 µg of IRBP_651−670 (LAQGAYRTAVDLESLAQT) (GeneCust) emulsified in Complete Freund’s Adjuvant (CFA) containing 3.3 mg/ml Mycobacterium tuberculosis H37Ra (Sigma-Aldrich & BD Difco). Mice also received an intraperitoneal injection of 1 µg of purified Bordetella pertussis toxin (Sigma-Aldrich) [31].

For adoptive transfer, donor mice were immunized as described above. 10 days after immunization, spleens and lymph nodes were collected and mechanically dissociated. Cell suspensions were cultured for 72 h in RPMI 1640 supplemented with fetal bovine serum 10%, gentamycin (50 µg/mL) and 2-mercaptoethanol (50 µM) in the presence of IRBP_651−670 (25 µg/mL) and IL-12 (5 ng/mL) (Peprotech) for Th1 polarization or IL-6 (20 ng/ml), IL-23 (10 ng/mL), IL-1β (10 ng/mL) and TGF-β (2 ng/mL) (Miltenyi) for Th17 polarization. After 3 days, lymphocytes were purified by gradient centrifugation on Ficoll-Paque Premium (GE Healthcare) and counted. 2 × 10⁷ cells were injected intravenously into recipient mice.

Clinical signs of EAU were assessed daily by fundus examination on a scale of 0–5 based on the extent of inflammation, as described in supplementary Fig. 6 [32].

Fig. 6.

P2X7 deficiency in MdMs affects the polarization of CD4⁺ T cells during EAU (A) IRBP-specific T lymphocytes from spleens and lymph nodes of WT mice were stimulated with IL12 (Th1 polarization), IL1β, IL6, IL23, and TGFβ (Th17 polarization), or IL6, IL23 and TGFβ (left panel). After 3 days, cells were collected and analyzed by flow cytometry. Representative plots of IFNγ and IL17A expression by CD4⁺ T cells. Bars represent the percentages of IFNγ⁺, IL17A⁺, or double positive (DP) CD4⁺ T cells under each stimulation condition (*p ≤ 0.05, **p ≤ 0.01 and ***p ≤ 0.001 compared to IL1β, IL6, IL23, and TGFβ stimulation). Data were analyzed by one-way ANOVA and are expressed as mean ± SEM. B EAU was induced in P2X7^+/+ and P2X7^−/− mice by adoptive transfer of IRBP-specific T lymphocytes from spleens and lymph nodes of WT mice polarized to Th17 with IL6, IL23, IL1β and TGFβ (left panel). At day 6 after adoptive transfer, cells from the retina were collected and analyzed by flow cytometry. Representative plots of IFNγ and IL17A expression by CD4⁺ T cells in the retinas of P2X7^+/+ and P2X7^−/− mice after adoptive transfer of Th17 polarized T cells. Bars represent the percentages of IFNγ⁺, IL17A⁺, or double positive (DP) CD4⁺ T cells in the retina 6 days after adoptive transfer. Each dot represents one retina Th17: n = 14 for P2X7^+/+ and n = 9 for P2X7^−/−, IL17A⁺ **p = 0.002 and DP **p = 0.002). C EAU was induced in P2X7^lox and Cx3cr1^creP2X7^lox mice by adoptive transfer of IRBP-specific T lymphocytes from the spleens and lymph nodes of WT mice polarized to Th17 with IL-6, IL-23, IL-1β and TGF-β (left panel). At day 6 after adoptive transfer, cells from the retina were collected and analyzed by flow cytometry. Representative plots of IFNγ and IL17A expression by CD4⁺ T cells in the retinas of P2X7^lox and Cx3cr1^creP2X7^lox mice after adoptive transfer of Th17 polarized T-cells. Bars represent the percentages of IFNγ⁺, IL17A⁺ or DP CD4⁺ T cells in the retina 6 days after adoptive transfer. Each dot represents one retina (n = 14 for P2X7^lox and n = 12 for Cx3cr1^creP2X7^lox, IL17A⁺ **p = 0.001 and DP *p = 0.013). D ELISA results show IL-1β release from peritoneal MdM and microglia from P2X7^+/+, P2X7^−/− (left panel), P2X7^lox and Cx3cr1^creP2X7^lox (right panel) mice stimulated for 3 h with LPS followed by ATP 30 min (MdM P2X7^+/+: 5765 ± 117 pg/mL, Microglia P2X7^+/+: 105.7 ± 20.8 pg/mL, MdM P2X7^lox: 4848 ± 943 pg/mL, Microglia P2X7^lox: 75.2 ± 14.8 pg/mL). ND: not detected. E Peritoneal MdM from P2X7^+/+ and P2X7^−/− mice were stimulated for 6 h with LPS followed by ATP. Then IRBP-specific T lymphocytes from spleens and lymph nodes of WT mice were added and stimulated with IL6, IL23 and TGFβ (left panel). After 3 days, cells were collected and analyzed by flow cytometry. Representative plots of IFNγ and IL17A expression by CD4⁺ T cells. Bars represent the percentages of IFNγ⁺, IL17A⁺, or double positive (DP) CD4⁺ T cells in presence of MdM from P2X7^+/+ or P2X7^−/− mice (*p = 0.05). Data were analyzed by Mann–Whitney tests and are expressed as mean ± SEM

Samples were collected 12 days after active EAU induction and 6 days after passive EAU, corresponding to the disease peak in our experimental models.

Flow cytometry

Mice were euthanized by CO₂ overdose followed by cervical dislocation. Lymph nodes were mechanically dissociated into a Petri dish and filtered through a 70 μm cell strainer to obtain a single cell suspension. Individual eyes were collected in PBS on ice and dissected under a binocular dissecting microscope by first sectioning the limbus and removing the cornea and lens. The retina was then carefully detached from the choroid using tweezers and transferred to a new 1.5 ml tube containing cold PBS (4 °C). Individual retinas were incubated at 37 °C for 30 min in PBS containing the enzyme Liberase TL at 1.6 Wunsch unit/mL (Sigma-Aldrich). Individual retinas were then filtered and rinsed through a 70 μm cell strainer to obtain a single cell suspension. Dead cells were stained with Viobility™ Fixable Dye (Miltenyi Biotec). Cytometry was performed using the following antibody clones specific for: CD45 (REA737), CD11b (REA592), Ly6C (REA796), Ly6G (REA526), MHCII (REA813), CD11c (REA754), CD3 (REA641), CD4 (REA604), CD44 (REA64), CD62L (REA828), CD80 (REA983), CD86 (REA1190) (Miltenyi Biotec) (Supplemental Table 1) and P2X7 (1F11) (Biolegend). For surface labeling, cells were incubated with the respective antibodies for 30 min at 4 °C and then fixed in paraformaldehyde 1% (PFA). For the FOXP3 detection, a FOXP3 staining kit and anti-Foxp3 antibody (REA788) (Miltenyi Biotec) were used according to the manufacturer’s instructions.

For intracellular labeling of IFN-γ and IL-17 A, cells were incubated for 4 h at 37 °C in RPMI with the cell stimulation and protein transport inhibitor cocktails (ebioscience™). After surface staining, cells were fixed and permeabilized with fixation/permeabilization buffer (Invitrogen) and stained with antibodies against IFN-γ (REA638) and IL-17 A (REA660). Acquisition was performed on a Celesta SORP cytometer (BD Biosciences), and data were analyzed with FlowJo 10.8.

Cell culture

Microglia were isolated with Percoll separation as previously described [33]. Briefly, mice were perfused with PBS, and brain hemispheres were dissected and homogenized in a digestion cocktail.

containing the enzyme Liberase TL at 1.6 Wünsch unit/mL and DNAse1 at 0.5 mg/mL (Sigma-Aldrich) for 30 min at 37 °C in 5% CO2. After washing, the brain homogenate was centrifuged in a 25/75% Percoll gradient (GE Healthcare). Microglia were recovered from the 25/75% Percoll interphase. Mouse peritoneal macrophages were isolated using a quick peritoneal lavage method with PBS. Cells were washed and seeded at a density of 50,000 cells/well in 96-well plates. Then, cells were stimulated directly for microglia or after 24 h for peritoneal MPs, with LPS 40 ng/mL (E. coli 0127:B8, Sigma) for 3 h, followed by ATP 5mM (Sigma-Aldrich) 30 min at 37 °C 5% CO₂. IL1β levels in the supernatant were assessed with mouse IL1β DuoSet ELISAs according to the manufacturer’s instructions (R&D Systems).

Fluorescence activated cell-sorting

Preparation of cell suspensions from individual mouse retinas and lymph nodes was performed as described for flow cytometry. Retinal and lymph node cells were labelled with the following antibodies: anti-CD45, anti-CD11b, anti-CD11c and anti-Ly6G (Miltenyi)(Supplemental Table 1). Dead cells were excluded from the analysis using a Live/Dead fixable staining reagent (Miltenyi). CD45⁺CD11b⁺Ly6G^neg retinal cells and CD45⁺CD11b⁺CD11c⁺ lymph node cells were sorted on the Melody cytometer (BD Biosciences), collected in a chilled tube, and then used for RNA and scRNA-sequencing.

Bulk and single-cell RNA sequencing

For bulk RNA sequencing: The SMART-Seq v4 Ultra Low Input RNA Kit for Sequencing (Takara) was used to generate cDNA from 3 ng of total RNA. RNAseq libraries were prepared from cDNA using the Nextera XT DNA Library Preparation Kit (Illumina) and sequenced on a Novaseq 6000 sequencer (Illumina).

Fastq files obtained from the sequencing were aligned using STAR (v2.7.9a) against the mouse reference genome from Ensemble Mouse v106 (2022/06/21), with option “--quantMode GeneCounts” to extract the raw counts for each gene, and all count files were concatenated into a single file. The sequencing depth ranged between 28 and 47 million aligned reads. A sample file was created with the sample data including the conditions (P2X7^lox and Cx3cr1^creP2X7^lox) and the replicate numbers. The count file and the sample file were loaded in our in-house R Shiny application EYE DV seq. We added 1 to all counts in the count file to avoid any 0 read count errors. We then removed the genes having a total count < 10 across all samples. Finally, the DESeq2 (v1.40.2) analysis was performed comparing the groups defined in the ‘condition’ column, with ‘P2X7^lox’ as control group and ‘Cx3cr1^creP2X7^lox’ as condition. The results were then filtered for significant genes using the padj-values < 0.05, BaseMean > 10, and Log2FC > 0.75. An enrichment analysis was performed to study the implicated pathways, using the Reactome database (using the Bioconductor package ‘ReactomePA’ (v1.44.0) and the GO database (using the Bioconductor packages org. Mm.eg.db (3.10.0) and ‘clusterProfiler’ (v 4.8.1)).

We performed the Gene Set Enrichment Analysis (GSEA) using the Broad Institute application (https://www.gsea-msigdb.org/gsea/index.jsp). We used the reference Mouse_Ensembl_Gene_ID_MSigDB.v2023.2.Mm.chip as chip platform, and the gene sets database m5.go.bp.v2023.2.Mm.symbols.gmt to which we added our custom gene sets for the inflammasome and phagocytosis pathways. We then ran the application with the default parameters. To test our gene sets’ significance, we used ROAST [34] from the Bioconductor package limma (https://bioconductor.org/packages/release/bioc/html/limma.html). We investigated our gene sets for inflammasome and chemokine pathways with 10,000 rotations as per their guidelines.

For single-cell RNA sequencing (scRNAseq): The barcoded cDNAs were prepared from sorted cells using a 10x genomics chromium and were used to construct the indexed scRNA sequencing libraries according to Chromium Next GEM Single Cell 3′ Library v3.1 and dual Index Kit TT according to the manufacturer’s protocol. Libraries were sequenced on an Illumina Novaseq 6000 sequencer. The Fastq files were processed using Cellranger v6.1.2 with default options (alignment, count). The resulting expression matrices were merged and analyzed with Seurat v4 (merge, QC filtering, normalization, PCA, clustering, UMAP/tSNE). The resulting Seurat object was used to create a H5AD file that was used as input for cellxgene for visualization and data exploration. Further analyses were performed in cellxgene using the cellxgene_VIP plugin (differential analysis, various visualizations, etc.). Processing scripts are available on demand. Genes interaction analysis was performed using STRING consortium [35].

Statistical analysis

Sample sizes for our experiments were determined according to our previous studies. Graph Pad Prism 10 (GraphPad Software) was used for data analysis and graphic representation. Data are presented as mean ± standard error of the mean (SEM). Differences between two groups were analyzed using an unpaired Mann-Whitney test and differences between multiple groups were analyzed using a one-way ANOVA. Significance levels are marked *P < 0.05; **P < 0.01; ***P < 0.001. The n and p-values are indicated in the figure legends.

Results

P2X7 deficiency decreases the severity of EAU

Studies of the role of P2X7 in experimental autoimmune encephalitis (EAE) have produced contradictory results [36–38] and earlier studies on EAU suggested that inhibition of P2X7 might reduce the severity of disease [39, 40], but the mechanism is unclear.

To evaluate whether P2X7 influenced EAU induced with the retinal autoantigen IRBP-peptide 651 to 670 (IRBP_651−670), the most reproducible EAU model, we first immunized P2X7^+/+ and P2X7^−/− mice with the retinal antigen in complete Freund’s adjuvant (CFA) (Fig. 1A). We chose to present the results at day 12 only, as this corresponds to the disease’s peak in our model and represents the phase in which the pathological differences are most pronounced. Clinical evaluation 12 days after the immunization revealed a significantly reduced disease in P2X7^−/− mice (clinical score = 1.6 ± 0.28) compared to P2X7^+/+ mice (clinical score = 3.2 ± 0.26). We used flow cytometry to analyze the immune cell infiltration in the retina during EAU. Our analysis of EAU retina at day 13 revealed a reduction of microglial cells (CD11b⁺CD45^med), monocyte-derived mononuclear phagocytes (MdMs) (CD11b⁺CD45^high), CD4⁺ T-cells (CD3⁺CD4⁺Foxp3^neg), and Tregs (CD3⁺CD4⁺Foxp3⁺) in P2X7^−/− mice (Fig. 1B). This reduction was significant for MdMs and CD4⁺ T-cells. Treg-mediated control of effector T-cells depended on the effector T-cells: Tregs ratio rather than absolute Tregs number. Despite the decrease in CD4⁺ T cells, the T-cells: Tregs ratio remained comparable in P2X7-deficient and WT mice, suggesting unaffected Treg expansion capacity during EAU in P2X7^−/− mice. The frequencies of immune cells in the draining lymph nodes were not affected by the lack of P2X7 expression (Fig. 1C). To identify the immune cells that potentially play a role in pathological processes through P2X7 expression, we analyzed P2X7 expression in immune cells 13 days after EAU induction. Flow cytometry analysis of the retina of naïve mice showed that all resident microglial cells expressed P2X7 (Fig. 1D). During EAU, microglial cells maintain their P2X7 expression without downregulation (95.1%±1.73) (Fig. 1E). Half of MdMs (51.2%±1.77) and CD4⁺ T cells (52.0%±2.54) and around 40% of Tregs (41.3%±1.54) that infiltrate the retina exhibit P2X7 expression (Fig. 1E).

Fig. 1.

P2X7 deficiency decreases the severity of EAU P2X7^+/+ and P2X7^−/− mice were immunized, or not (naïve), with 200 µg IRBP_651−670 peptide and injected with 1 µg Pertussis Toxin (PTX). A Representative fundus images and clinical scores of P2X7^+/+ and P2X7^−/− mice 12 days after immunization. Each dot represents one mouse (n = 9 for P2X7^+/+ and n = 7 for P2X7^−/−, **p = 0.005). The data shown are representative of four independent experiments. B Bars represent the percentages of immune cells in the retina: CD11b⁺CD45^med (Microglia), CD11b⁺CD45^high (monocyte-derived macrophages (MdMs)), CD3⁺CD4⁺Foxp3^neg (CD4⁺ T cells) and CD3⁺CD4⁺Foxp3⁺ (Tregs), 13 days after immunization. Each dot represents one retina (n = 5 for P2X7^+/+ and n = 6 for P2X7^−/−, CD11b⁺CD45^high *p = 0.030, CD3⁺CD4⁺Foxp3^neg *p = 0.035). C Bars represent the percentages of immune cells in the cervical lymph nodes 13 days after immunization. Each dot represents one mouse (n = 10 for P2X7^+/+ and n = 11 for P2X7^−/−). Data were analyzed by Mann–Whitney test and are expressed as mean ± SEM. D Cells from the retinas of P2X7^+/+ and P2X7^−/− naïve mice were analyzed by flow cytometry. Live cells were gated on CD11b⁺CD45^med (Microglia) (left panel) and a representative plot of P2X7 expression is shown (right panel). E After 13 days, cells from the retinas of P2X7^+/+ and P2X7^−/− immunized mice were analyzed by flow cytometry. Live cells were gated on CD11b⁺CD45^med (Microglia), CD11b⁺CD45^high (MdMs), CD3⁺CD4⁺Foxp3^neg (CD4⁺ T cells), and CD3⁺CD4⁺Foxp3⁺ (Tregs) to evaluate P2X7 expression on each subset. Representative plots of gating and P2X7 expression in the retina of EAU mice are shown (left panel). The percentage of P2X7 + cells for each cell subset is shown in a bar graph (right panel), each dot represents one retina (n = 16 for CD11b⁺ cells and n = 11 for CD4⁺ cells)

These findings confirm that P2X7 plays a significant role in the overall development of EAU. We show that both innate and adaptive immune cells express P2X7, prompting the question of how P2X7 influences the various functions of different immune cells during EAU.

P2X7-deficiency does not affect the activation of dendritic cells and CD4⁺ T cells

The decreased disease severity observed in P2X7-deficient mice could be attributed to a reduced number of autoreactive T-cells resulting from impaired T cell activation. Several studies suggest a role of P2X7 during antigen presentation and T-cell activation [41]. Next, we investigated whether the absence of P2X7 might affect T-cell activation by antigen-presenting cells. We analyzed dendritic cells and T cell activation in lymph nodes of P2X7^+/+ and P2X7^−/− mice by flow cytometry, 10 days after immunization with the IRBP peptide.

We first analyzed P2X7 expression in dendritic cells (MHCII⁺CD11c⁺) and noticed two subpopulations: one expressing P2X7 (MHCII⁺CD11c⁺CD11b⁺; P2X7 mean expression 1412 ± 69.5) and the other not expressing it (MHCII⁺CD11c⁺CD11b⁻; P2X7 mean expression 182.3 ± 2.8) (Fig. 2A). To activate T-cells, dendritic cells present the antigen on the surface of MHC and provide a co-stimulation signal via the expression of CD80 and CD86. The CD11c⁺CD11b⁺ population expressing P2X7 expressed the co-stimulatory molecules CD80 and CD86, but the absence of P2X7 did not affect their expression levels (Fig. 2B, C). We also analyzed the transcriptome of FACS sorted CD11c⁺CD11b⁺ cells from the lymph nodes of immunized P2X7^+/+ and P2X7^−/− mice using RNA sequencing and did not observe any major differences (Suppl Fig. 1). These findings suggested that lack of P2X7 did not affect the activation state of DCs in the lymph nodes.

Fig. 2.

P2X7 deficiency does not affect dendritic cell (DC) and CD4⁺ T cell activation P2X7^+/+ and P2X7^−/− mice were immunized with 200 µg IRBP_651−670 peptide. Inguinal lymph nodes were collected 10 days post-immunization. A DCs were defined as MHCII⁺CD11c⁺ (left panel). Representative plots of P2X7 and CD11b expression for MHCII⁺CD11c⁺ cells are shown (middle panel). Representative plots of P2X7 expression for each subset MHCII⁺CD11c⁺CD11b^+/− are shown (right panel). Representative plots of CD80 B and CD86 C expression for each subset are shown (left panel). Quantification of the expression of co-stimulatory molecules CD80 (B) and CD86 (CD) for each subset is shown (right panel), each dot represents pooled cells from 3 mice (n = 3 for P2X7^+/+ and P2X7^−/−). D Representative plots of P2X7 expression by CD3⁺CD4⁺Foxp3^neg (CD4⁺ T cells) and CD3⁺CD4⁺Foxp3⁺ (Regulatory T cells) in the lymph nodes (left panel). The percentage of P2X7 + cells for each subset is shown (right panel), each dot represents one mouse (n = 10). (E-F) Gating of P2X7 + cells for subsets of T cells (right panels). Bars represent the percentages of CD3⁺CD4⁺ T cells E and CD4⁺Foxp3⁺ T cells F in the lymph nodes, each dot represents one mouse (n = 5 for P2X7^+/+ and n = 4 for P2X7^−/−). Expression of cell activation markers by CD4⁺ T cells was evaluated by flow cytometry. CD4⁺ T cells were gated on CD44⁻CD62l⁺ or CD44⁻CD45RB⁺ (naïve T cells), CD44⁺CD62l⁺ or CD44⁺CD45RB⁺ (central memory T cells), and CD44⁺CD62l⁻ or CD44⁺CD45RB⁻ (effector T cells) (G and H). Each dot represents represent pool cells from 2–3 mice (n = 3–4 for P2X7^+/+ and n = 5 for P2X7^−/− mice). Data were analyzed by multiple Mann–Whitney tests and are expressed as mean ± SEM

On the other hand, we observed that only a proportion of CD4⁺Foxp3^neg T cells (17.9%±0.86; P2X7 mean expression 269.1 ± 11.2) expressed P2X7, while the majority of Tregs (CD4⁺Foxp3⁺) (82.2%±3.60; P2X7 mean expression 1346.0 ± 29.9) expressed P2X7 (Fig. 2D). Furthermore, we did not observe any significant effect in the frequencies of CD4⁺ T cells and CD4⁺Foxp3⁺ T cells associated with the lack of P2X7 expression (Fig. 2E, F). CD4⁺ T cell activation states were evaluated by the cell surface expression of the adhesion proteins CD44 and CD62l and the enzyme CD45RB: naïve T cells (CD44⁻CD62l⁺ or CD44⁻CD45RB⁺), central memory T cells (CD44⁺CD62l⁺ or CD44⁺CD45RB⁺) and effector T cells (CD44⁺CD62l⁻ or CD44⁺CD45RB⁻). We also analyzed the cell surface marker CD45RB because it has been shown that P2X7 activation could lead to cleavage of CD62l [42]. The proportion of each population of CD4⁺ T cells were not significantly different between P2X7^+/+ and P2X7^−/− mice (Fig. 2G, H).

Altogether, these results suggested that lack of P2X7 did not impair DC and CD4⁺ T cell activation, making it unlikely that reduced generation of autoreactive T cells accounts for the decreased retinal infiltration observed in P2X7-deficient mice during EAU.

Lack of P2X7 expression in CD4⁺ T cells does not alter their functions during EAU

Based on our observations, P2X7 deficiency did not prevent the generation of autoreactive T cells. However, absence of P2X7 in T cells could impact their ability to cross the blood retinal barrier via the shedding of CD62l and also affect their pathogenicity via the release of cytokines in situ. Furthermore, some effector CD4⁺ T cells but also the majority of Tregs expressed P2X7 in the lymph nodes (Fig. 2D). Tregs were shown to be highly sensitive to cell death induced by P2X7 activation [43, 44]. Thus, we asked whether P2X7 deletion in Tregs could influence their ability to regulate effector T cells in vivo during EAU.

To answer this question, we first compared EAU induced by adoptive transfer of activated T cells isolated from P2X7^+/+ or P2X7^−/− donor mice to wild-type (WT) recipient mice (Fig. 3A). All mice developed EAU with similar clinical score 4 days after adoptive transfer of P2X7^+/+ (clinical score = 3.9 ± 0.10) or P2X7^−/− (clinical score = 3.9 ± 0.07) T cells (Fig. 3B). We also analyzed P2X7 expression by CD4⁺Foxp3^neg and CD4⁺Foxp3⁺ T cells in the retina after adoptive transfer. Interestingly, in the retinas of WT mice that received P2X7^−/− T cells we observed similar number of CD4⁺Foxp3⁺ T cells expressing P2X7 as in mice receiving P2X7^+/+ T cells, suggesting that infiltrating Tregs predominantly originated from the recipient WT mice in this model (Fig. 3C). Then, we analyzed immune cell infiltration in the retina and the frequency of T cells in cervical draining lymph nodes by flow cytometry 6 days after adoptive transfer. We showed that the number of microglia, infiltrating MdM, and CD4⁺Foxp3^neg and CD4⁺Foxp3⁺ T cells were similar in the retina of mice receiving P2X7^+/+ or P2X7^−/− T cells as well as the frequency of Tregs among CD4⁺ T cells (Fig. 3D). In addition, no significant difference was observed in the lymph nodes (Fig. 3E). These results suggested that the absence of P2X7 expression by T cells did not affect their ability to cross the blood retinal barrier and to induce EAU.

Fig. 3.

Lack of P2X7 expression in CD4⁺ T cells does not alter their functions during EAU (A) EAU was induced in WT mice by adoptive transfer of IRBP-specific T lymphocytes from spleens and lymph nodes of P2X7^+/+ and P2X7^−/− mice. B Representative fundus images and clinical scores of WT recipient mice 4 days after adoptive transfer of P2X7^+/+ and P2X7^−/− T cells. Each dot represents one mouse (n = 10 for P2X7^+/+ and n = 9 for P2X7^−/−). The data shown are representative of three independent experiments. C Representative plots of P2X7 expression by CD3⁺CD4⁺Foxp3^neg (CD4⁺ T cells) and CD3⁺CD4⁺Foxp3⁺ (Regulatory T cells) in the retinas of WT mice after adoptive transfer of P2X7^+/+ or P2X7^−/− T cells. D Bars represent the percentages of immune cells in the retina 6 days after adoptive transfer. Each dot represents one retina (n = 10 for P2X7^+/+ and n = 8 for P2X7^−/−). E Bars represent the percentages of immune cells in the cervical lymph nodes, 6 days after adoptive transfer. Each dot represents one mouse (n = 5 for P2X7^+/+ and n = 4 for P2X7^−/−) (F) P2X7^lox and Foxp3^creERP2X7^lox mice were immunized with 200 µg IRBP_651−670 peptide and injected with 1 µg PTX, one week after the end of tamoxifen treatment. G Representative fundus images and clinical scores of P2X7^lox and Foxp3^creERP2X7^lox 12 days after immunization. Each dot represents one mouse (n = 13 for P2X7^lox and n = 8 for Foxp3^creERP2X7^ox). The data shown are representative of four independent experiments. H Representative plots of P2X7 expression by CD3⁺CD4⁺Foxp3^neg (CD4⁺ T cells) and CD3⁺CD4⁺Foxp3⁺ (Regulatory T cells) in the retinas of P2X7^lox and Foxp3^creERP2X7^lox mice 13 days after immunization. I Bars represent the percentages of immune cells in the retina 13 days after immunization. Each dot represents one retina (n = 26 for P2X7^lox and n = 16 for Foxp3^creERP2X7^lox). J Bars represent the percentages of immune cells in the cervical lymph nodes, 13 days after immunization. Each dot represents one mouse (n = 13 for P2X7^lox and n = 8 for Foxp3^creERP2X7^lox). Data were analyzed by Mann–Whitney test and are expressed as mean ± SEM

In EAU experiments induced by adoptive transfer of T cells we noted that the vast majority of effector T cells appeared to come from the donor mice, whereas regulatory T cells seemed to originate from the recipient mice. Therefore, we conducted experiments transferring T cells from CD45.1 mice into CD45.2 mice and demonstrated that ~ 77% of CD4⁺Foxp3^neg T cells express the donor marker CD45.1, compared to only ~ 13% for CD4⁺Foxp3⁺ T cells (Suppl Fig. 2). These results are consistent with our findings showing P2X7 expression in Tregs after transfer of activated P2X7-deficient T cells (Fig. 3C).

Then, to investigate the specific role of P2X7 expressed by Tregs independently of its expression by effector T cells, we generated mice deficient for P2X7 in Tregs (Foxp3^creERxP2X7^lox mice). One week after tamoxifen treatment, we induced active EAU in P2X7^lox and Foxp3^creERxP2X7^lox mice (Fig. 3F). We evaluated the clinical score by fundoscopic examination and showed that the severity of the disease was not significantly different between P2X7^lox (clinical score = 0.9 ± 0.20) and Foxp3^creERP2X7^lox (clinical score = 1 ± 0.25) mice 12 days after EAU induction (Fig. 3G). Flow cytometry analysis of T-cells from the retina of Foxp3^creERxP2X7^lox mice during EAU showed expression of P2X7 by CD4⁺Foxp3^neg T cells but not CD4⁺Foxp3⁺ T cells confirming the specific deletion of P2X7 in Tregs (Fig. 3H). Flow cytometry analysis of retinal inflammation showed that P2X7 deletion in Tregs did not affect the infiltration of macrophages and CD4⁺Foxp3⁺ T cells in the retina, 12 days after EAU induction (Fig. 3I, J). Interestingly, neither the frequency of CD4⁺Foxp3^neg T cells nor the proportion of Tregs among CD4⁺ T cells in the retina and in the cervical lymph nodes were affected by P2X7 deletion (Fig. 3I, J). These results suggested that the specific deletion of P2X7 in Tregs did not impair the auto-immune response.

Taken together, our data suggested that expression of P2X7 in T cells did not play a detectable role in the development of EAU.

P2X7-deficiency in resident microglial cells and infiltrating MdMs reduces the severity of EAU

Since P2X7 expression on lymphocytes does not seem to play a major role in EAU development, and given that its activation in macrophages induces a pro-inflammatory response, we hypothesize that P2X7 primarily influences EAU pathogenesis through its expression on resident microglia and infiltrating MdMs. Thus, we investigated the role of P2X7 expressed in innate immune cells during EAU. The disease was induced in Cx3cr1^creP2X7^lox mice in which the P2rx7 gene is specifically deleted in MdMs and microglia compared to P2X7^lox control mice (Fig. 4A). Clinical assessment of these animals by fundoscopic examination showed that the severity of EAU was significantly attenuated in Cx3cr1^creP2X7^lox mice (clinical score = 1.3 ± 0.3) vs. P2X7^lox mice (clinical score = 2.2 ± 0.3) (Fig. 4B). P2X7 deletion in microglia (P2X7 Mean expression in P2X7^lox: 3884.9 ± 147.9 and Cx3cr1^creP2X7^lox: 805.8 ± 26.2) and MdMs (P2X7 Mean expression in P2X7^ox: 1472.0 ± 78.4 and Cx3cr1^creP2X7^lox: 401.2 ± 52.0) in the retina of Cx3cr1^creP2X7^lox mice was confirmed by flow cytometry (Fig. 4C). Flow cytometry analysis of immune cells in the retina showed that lack of P2X7 in microglia and MdMs resulted in a significant decrease in microglia, MdMs, CD4⁺ T-cells and Tregs, with no effect on the frequency of Tregs among CD4⁺ T cells (Fig. 4D). No significant variation in the frequencies of immune cells between both groups was noticed in the cervical lymph nodes (Fig. 4E).

Fig. 4.

P2X7 deficiency in resident microglial cells and infiltrating MdMs reduces the severity of EAU (A) P2X7^lox and Cx3cr1^creP2X7^lox mice were immunized with 200 µg IRBP651-670 peptide and injected with 1 µg Pertussis Toxin (PTX). B Representative fundus images and clinical scores of P2X7^lox and Cx3cr1^creP2X7^lox 12 days after immunization. Each dot represents one mouse (n = 16 for P2X7^lox and n = 12 for Cx3cr1^creP2X7^lox mice, *p = 0.044). The data shown are representative of at least five independent experiments. C Representative plots of P2X7 expression by CD11b⁺CD45^med (Microglia) and CD11b⁺CD45^high (Infiltrating MdM) in the retinas of P2X7^lox and Cx3cr1^creP2X7^lox mice 13 days after immunization. D Bars represent the percentages of immune cells in the retina 13 days after immunization. Each dot represents one retina (P2X7^lox (n = 32) vs. Cx3cr1^creP2X7^lox mice (n = 24); CD11b⁺CD45^med **p = 0.006, CD11b⁺CD45^high *p = 0.011, CD3⁺CD4⁺Foxp3^neg **p = 0.003 and CD3⁺CD4⁺Foxp3⁺ **p = 0.006). (E) Bars represent the percentages of immune cells in the cervical lymph nodes, 13 days after immunization. Each dot represents one mouse (n = 16 for P2X7^lox and n = 12 for Cx3cr1^creP2X7^lox mice). Data were analyzed by Mann–Whitney test and are expressed as mean ± SEM

To identify the molecular pathways affected by the P2X7-deficiency, we cell-sorted CD45⁺CD11b⁺ cells by fluorescence-activated cell sorting from the retinas of Cx3cr1^creP2X7^lox and P2X7^lox mice 13 days after EAU induction and analyzed their transcriptomes by RNAsequencing (RNAseq). We found that 1287 genes were significantly down-regulated and 1086 genes were up-regulated in Cx3cr1^creP2X7^lox compared to P2X7^lox (Log₂FC > 0.75 and padj < 0.05; Suppl Tabular data 1). We primarily focused our analysis on down-regulated genes because we hypothesized that the absence of P2X7 improved the pathology by reducing the inflammatory response (Fig. 5A). Interestingly, among the genes that were the most regulated we found Adam10 and Adam17, two genes involved in the release of TNFα, as well as genes that belong to the complement pathway (C3, C4b, Cfh) and a large number of genes belonging to the phagocytosis pathway (48 genes of the 233 genes that belong to the phagocytosis pathway Goterm GO0006909) (Fig. 5A, B). Indeed, when we carried out gene set enrichment analysis (GSEA) we found a significant enrichment in phagocytosis-associated genes (normalized enrichment score (NES) = 1.47, p = 0.0, false discovery rate (FDRq) = 0.09) (Fig. 5B). In addition, we examined whether the absence of P2X7 could lead to modification of the expression of other purinergic receptors (P2X and P1 receptors) or ATP-degrading enzymes (CD39 and CD73 ectonucleotidases). We observed that the levels of the expression of these genes were not significantly different between cells from Cx3cr1^creP2X7^lox and P2X7^lox mice (Fig. 5C). While individual genes may not exhibit statistically significant modulation, the collective group of genes to which they belong can undergo significant modulation. Thus, we also conducted a complementary analysis of the candidate pathways. We selected genes of the inflammasome and chemokine pathways that were previously associated with P2X7 activation. We found that the inflammasome pathway was significantly down-regulated in CD11b⁺ cells from Cx3cr1^creP2X7^lox (FDR = 0.001) mice but not genes of the chemokine family (Fig. 5D). We also noticed a significant GSEA enrichment in transcripts related to the inflammasome (NES = 1.55; p = 0.04, FDRq = 0.07).

Fig. 5.

P2X7 deficiency in resident microglial cells and infiltrating MdMs decreases the expression of inflammatory-related genes during EAU (A) Volcano plot showing significantly differentially regulated genes between CD45⁺CD11b⁺ cells from the retina of P2X7^lox (n = 4) and Cx3cr1^creP2X7^lox mice (n = 4) 13 days after immunization. B Heatmap and GSEA cluster of genes belonging to the phagocytosis pathway (Goterm GO0006909) in CD45⁺CD11b⁺ cells from the retina of P2X7^lox (n = 4) and Cx3cr1^creP2X7^lox mice (n = 4) developing EAU. C Bars showing the expression of P2x and P1 receptors, and ectonucleotidases in CD45⁺CD11b⁺ cells from the retina of P2X7^lox (n = 4) and Cx3cr1^creP2X7^lox mice (n = 4). D Heatmap and GSEA cluster of genes belonging to the inflammasome pathway (left panel) and chemokine family (right panel) in CD45⁺CD11b⁺ cells from the retina of P2X7^lox (n = 4) and Cx3cr1^creP2X7^lox mice (n = 4) developing EAU

Altogether, these results suggested that the lack of P2X7 expression by MPs decreased the severity of the EAU and could be attributed to the down regulation of complement, inflammasome and phagocytosis pathways.

P2X7 deficiency in resident microglial cells and infiltrating MdMs reduces Th17 cell differentiation in vivo

Th17 cells play a major effector function in EAU pathogenesis and the IL17 response is associated with disease severity [45]. Previous studies have indicated that total knockout of P2X7 can alter the differentiation of CD4⁺ T cells into Th1 and Th17 subsets [2, 11], and our observations revealed that the lack of P2X7 in MPs downregulated the inflammasome pathway, which has been implicated in Th17 polarization [46]. So, we questioned whether the absence of P2X7 in microglia and MdMs could influence the pathogenic T-cell response.

First, when we induced Th17 polarization in activated T cells by treating them in vitro with IL6, IL23, IL1β, and TGFβ, we observed a decrease in IFNγ⁺ CD4⁺T cells but an increase in the number of IL17A–producing CD4⁺ T cells and IL17A⁺IFNγ⁺ double positive (DP) cells (increase 133.7% and 297.3% respectively) compared to IL12 treatment. The lack of IL1β in the cytokine cocktail resulted in a reduced number of IL17A–producing CD4⁺ T cells and double positive (DP) cells (decrease 32.0% and 42.3% respectively) underscoring the critical role of IL1β in promoting IL17A expression in T cells (Fig. 6A).

To investigate the role of P2X7 in Th17 T-cell polarization, we intravenously transferred activated, Th17-polarized T cells into P2X7^+/+ or P2X7^−/− recipient mice. Six days post-transfer, we analyzed the frequencies of intra-retinal CD4⁺ T cells producing IFNγ and/or IL-17 A (Fig. 6B), which primarily originated from the donor cells (see Supplemental Fig. 2). We observed no significant differences in the proportion of IFNγ-producing CD4⁺ T-cells while the proportion of IL17-producing CD4⁺ T-cells and double positive (DP) cells were significantly decreased in P2X7^−/− (decrease 43.7% and 43.2% respectively) (Fig. 6B). To demonstrate the involvement of P2X7 expressed by microglia and MdMs in the observed shift in T-cell polarization, we adoptively transferred Th17 polarized T-cells in Cx3cr1^creP2X7^lox and P2X7^lox mice. We showed a reduced frequency of IL17A⁺ and DP CD4⁺ T-cells in Cx3cr1^creP2X7^lox mice compared to P2X7^lox mice (decrease 49.1% and 41.2% respectively) (Fig. 6C). Since IL1β is a critical mediator of Th17 responses, we next assessed its release by ATP-stimulated MdMs and microglia. Remarkably, MdMs produced at least 50 times more IL1β than microglia (MdM P2X7^+/+: 5765 ± 117 pg/mL, Microglia P2X7^+/+: 105.7 ± 20.8 pg/mL) (Fig. 6D), suggesting that MdMs are the main regulators of T-cell polarization. Additionally, experiments using cells from P2X7^-/- and Cx3cr1^creP2X7^lox mice showed no detectable IL1β levels in the supernatants under the same conditions as the control mice, confirming the functional invalidation of P2X7 in MPs from Cx3cr1^creP2X7^lox mice (Fig. 6D). To directly link P2X7-dependent IL1β release to Th17 polarization, we performed co-cultures of WT T cells with either WT or P2X7-deficient MdMs activated by ATP (Fig. 6E). Consistent with our in vivo data, we showed that the number of IL17A-expressing T-cells was reduced in the presence of P2X7^-/- MdMs compared to WT MdMs (decrease 36.1% and 100.4% of IL17A–producing and DP CD4⁺ T cells respectively).

Together, these results suggested that P2X7 expression in MPs plays a pathogenic role during EAU in vivo by promoting Th17 T cells, known to play an important role in the pathogenesis.

Lack of P2X7 affects microglia and infiltrating MdM subpopulations differently during EAU

We highlighted the different pathways affected by the absence of P2X7 expression in MdMs and microglia. To go further in our analysis, we investigated whether P2X7-deficiency had a different impact on the different MP subpopulations.

We performed scRNAseq on 27,712 sorted CD45⁺CD11b⁺LyG⁻ cells from the retina of naïve WT mice (Fig. 7A left), and P2X7^lox control mice and Cx3cr1^creP2X7^lox mice (P2X7-deficiency in all MPs) that clinically developed EAU as evaluated by fundoscopy (Fig. 7A right and Suppl Fig. 3). Unsupervised clustering revealed 9 major CD11b⁺ subpopulations in the retina based on the expression of 23 known cell-type-specific markers (Fig. 7A, B) [47, 48] and we observed that P2X7 was mainly expressed in microglia, MdM#1, and MdM#2 subpopulations (Fig. 7C). We identified the differentially expressed genes (DEGs) in these clusters between Cx3cr1^creP2X7^lox and P2X7^lox mice and found 22 genes significantly down-regulated and 6 genes up-regulated in EAU microglia, 62 down-regulated genes and 15 up-regulated genes in MdM#1 and 61 down-regulated genes and 10 up-regulated genes in MdM#2 (Log2FC > 0.25 and padj < 0.05; Suppl Tabular data 2). Interestingly, examination of downregulated DEGs revealed their association with different pathways in the three subpopulations (Fig. 7E, F). Cx3cr1, Cxcr4, and genes associated with the IFN pathway were modulated in activated microglia, whereas genes belonging to complement and phagocytosis pathways were exclusively impacted in MdM#1 and MdM#2 subpopulations. These data showed that lack of P2X7 modified the expression of different genes and pathways in activated microglia and infiltrating MdMs.

Fig. 7.

Microglia and infiltrating MdM subpopulations in the retina are differentially modified by lack of P2X7 during EAU Single-cell RNA sequencing analysis of CD11b⁺ cells from the retina of naive WT mice, P2X7^lox and Cx3cr1^creP2X7^lox mice 13 days after immunization. A UMAP representation of CD11b⁺ cell populations from the retina of naive WT mice and P2X7^lox mice during EAU. B CD11b⁺ cell population marker (x axis) expression levels for the 9 CD11b⁺ cell populations (y axis). Size of dots represents the fraction of cells expressing the marker, and color intensity indicates mean normalized scaled expression levels. C Feature plot showing P2X7 expression projected onto the UMAP representation of CD11b⁺ cell populations from the retinas of naive WT mice and P2X7^lox mice during EAU. D Volcano plot showing significantly differentially regulated genes in microglia, MdM#1, and MdM#2 subpopulations from P2X7^lox vs. Cx3cr1^creP2X7^lox mice developing EAU. E Full STRING Network of genes differentially expressed in CD11b⁺ cells from the retinas of P2X7^lox vs. Cx3cr1^creP2X7^lox cells in clusters microglia, MdM#1, and MdM#2 clusters. The edges indicate both functional and physical protein associations. Line thickness indicates the strength. F Dot plots showing normalized transcript expression distribution on a per cluster basis for genes that were differentially expressed in CD11b⁺ cells from the retinas of P2X7^lox vs. Cx3cr1^creP2X7^lox mice. Size of dots represents the fraction of cells expressing the marker, and color intensity indicates mean normalized scaled expression levels. Significantly differentially regulated genes are in bold. G Dot plot showing homeostatic, neurodegenerative, and IFN-responsive associated gene (x-axis) expression levels for naive and EAU microglial populations (y axis). Size of dots represents the fraction of cells expressing the gene, and color intensity indicates mean normalized scaled expression levels. H Feature plots showing microglial cells expressing the IFN-responsive transcriptional signature (Ifit3, Oas3, Ifitm3, Isg15, Irf7 and Stat1) projected onto the UMAP representation of the microglial population from the retina of P2X7^lox (left) and Cx3cr1^creP2X7^lox (right) mice during EAU

We further analyzed the microglial population in EAU and investigated the expression of genes linked with a homeostatic phenotype (C1qa, Csf1r, Hexb, Tmem119, P2ry12, Tgfbr1, Mef2a, Olfml3), neurodegeneration-associated signature genes (Apoe, Trem2, Cst7, Itgax, Spp1, Lpl, Fgf1) [49, 50], and IFN-response genes (Ifitm3, Isg15, Irf7, Stat1) [51, 52](Fig. 7G). We observed that EAU microglia expressed the core microglial markers C1qa, Csf1r, and Hexb. Consistent with previous findings in the EAU model, we found that TGFβ-dependent homeostatic genes (Tmem119, P2ry12, Tgfbr1, Mef2a, Olfml3) were down regulated [50]. Genes associated with mouse models of neurodegenerative diseases were barely expressed, such as Cst7, Itgax, Spp1, Lpl, and Fgf1, or were notably down-regulated, such as Trem2, a well-known marker for disease associated microglia (DAM) [53]. Interestingly, we showed heightened expression of the inflammatory associated gene Apoe and upregulation of the IFN-responsive markers Ifitm3, Isg15, Irf7, and Stat1 [51]. “IFN-response microglia” were initially identified in the central nervous system of mice in response to virus, LPS, and glioma [51] but also found after spinal cord injury [52]. We asked whether lack of P2X7 in MPs could affect this specific subpopulation. We assessed the frequency of “IFN-response microglia” defined by Ifit3, Oas3, Ifitm3, Isg15, Irf7, and Stat1 expression [51, 52] and found that this sub-population was decreased by ∼37% among microglia from the retinas of Cx3cr1^creP2X7^lox mice (292/2052 microglia − 14.2%) compared to P2X7^lox mice (288/1267 microglia − 22.7%) during EAU (Fig. 7H).

These results indicate that the absence of P2X7 differentially impacted the MP subpopulations, leading to a decrease in complement and phagocytosis pathways in MdMs and a reduction in the IFN-responsive microglia population.

P2X7 in microglia is involved in the pathogenic processes during EAU

Although P2X7 expression in MdMs plays an important role in uveitis through promoting Th17 T cell populations and other mechanisms, microglia have been reported to also play an important role in the development of EAU [54, 55]. This prompted us to specifically investigate whether P2X7 expression in microglia contributes to potentially detrimental microglial functions.

We used Cx3cr1^creERP2X7^lox mice to delete P2X7 expression specifically in microglial cells. Four weeks after tamoxifen treatment, we induced active EAU in P2X7^lox and Cx3cr1^creERP2X7^lox mice (Fig. 8A). Clinical assessment of these animals by fundoscopic examination showed that EAU severity was significantly reduced in Cx3cr1^creERP2X7^lox mice (clinical score = 1.2 ± 0.3) compared to P2X7^lox mice (clinical score = 1.9 ± 0.2) (Fig. 8B). We confirmed by flow cytometry that P2X7 was not present in microglia (P2X7 Mean expression in P2X7^lox: 3667.0 ± 268.9 and Cx3cr1^creERP2X7^lox: 1327.4 ± 51.2) but still expressed in infiltrating MdMs in Cx3cr1^creERP2X7^lox mice (P2X7 Mean expression in P2X7^lox: 2149.6 ± 114.9 and Cx3cr1^creERP2X7^lox: 2135.2 ± 221.1) during EAU (Fig. 8C). Analysis of microglia, infiltrating MdMs, and T-cells showed that the proportion of immune cells was significantly reduced in the retina of Cx3cr1^creERP2X7^lox mice vs. P2X7^lox mice during EAU (CD11b⁺CD45^med: P2X7^lox 0.36 ± 0.09% and Cx3cr1^creERP2X7^lox 0.13 ± 0.02%; CD11b⁺CD45^high: P2X7^ox 1.02 ± 0.42% and Cx3cr1^creERP2X7^lox 0.30 ± 0.12%; CD4⁺Foxp3^neg: P2X7^lox 1.19 ± 0.40% and Cx3cr1^creERP2X7^lox 0.26 ± 0.13%) (Fig. 8D). We did not observe any significant difference in the proportion of immune cells in the cervical lymph nodes between both groups of mice (Fig. 8E).

Fig. 8.

P2X7 deficiency on resident microglia is sufficient to reduce the severity of EAU (A) P2X7^lox and Cx3cr1^creERP2X7^lox mice were immunized with 200 µg IRBP_651−670 peptide and injected with 1 µg Pertussis Toxin (PTX). B Representative fundus images and clinical scores of P2X7^lox and Cx3cr1^creP2X7^lox 12 days after immunization. Each dot represents one mouse (n = 14 for P2X7^lox and n = 13 for Cx3cr1^creERP2X7^lox mice, *p = 0.042). The data shown are representative of three independent experiments. C Representative plots of P2X7 expression by CD11b⁺CD45^med (Microglia) and CD11b⁺CD45^high (Infiltrating MdM) in the retinas of P2X7^lox and Cx3cr1^creERP2X7^lox mice 13 days after immunization. D Bars represent the percentages of immune cells in the retina 13 days after immunization. Each dot represents one retina (P2X7^lox (n = 12) vs. Cx3cr1^creERP2X7^lox mice (n = 16); CD11b⁺CD45^med **p = 0.009, CD11b⁺CD45^high *p = 0.032, CD4⁺Foxp3^neg **p = 0.005 and CD4⁺Foxp3⁺ *p = 0.019). E Bars represent the percentages of immune cells in the cervical lymph nodes, 13 days after immunization. Each dot represents one mouse (n = 6 for P2X7^lox and n = 8 for Cx3cr1^creERP2X7^lox mice). Data were analyzed by Mann–Whitney test and are expressed as mean ± SEM

Overall, these data suggested that P2X7 expressed specifically by microglia also significantly contributes to the development of EAU.

Discussion

In this study, we analyzed the role of P2X7-expressing immune cell types in the development of EAU. We showed a pivotal role of P2X7-expressing MPs in the control of autoimmune neuroinflammation. We demonstrated that the absence of P2X7 in these cells leads to a down-regulation of the expression of pro-inflammatory molecules, including the NLRP3 inflammasome and complement elements. While the suppression of P2X7 on T cells had no detectable effect on the disease, MP P2X7 indirectly shaped the phenotype of Th17 cells. In addition, the lack of P2X7 in MPs significantly altered the subpopulation of IFN-responsive microglia that had a detrimental impact on the progression of EAU. Together, our data support a model in which P2X7 activation on MPs controls the phenotype of autoreactive T-cells, in addition to promoting pro-inflammatory processes within the P2X7-expressing MPs (Supplemental Fig. 8).

In agreement with previous studies [39, 56], we showed that ubiquitous lack of P2X7 during EAU protected against the clinical development of severe EAU and also decreased neuroinflammation in the retina (Fig. 1). Interestingly, although the persistence of the P2X7 expression in T cells in one of the P2X7 knockout mouse lines has been proposed as a potential explanation for the divergent outcomes observed in these mice during EAE [11, 15], we here showed that lack of P2X7 expression in CD4⁺ T cells did not significantly impact their activation and pathogenicity in EAU (Figs. 2 and 3). Tregs are particularly susceptible to P2X7-mediated apoptosis [43], suggesting that lack of P2X7 expression in these cells could skew the CD4⁺ T cell/Treg balance towards Tregs, thereby potentially mitigating disease severity. To test this hypothesis, we specifically ablated P2X7 in Tregs and we observed that P2X7 expression in these cells did not impact the frequency of effector T cells and Tregs in the retina. This is possibly due to the fact that although approximately 80% of Tregs expressed P2X7 in the draining lymph nodes (Fig. 2D), the level of ATP might not be sufficiently high to affect P2X7-dependent CD4⁺ T cell properties. In the diseased retina on the other hand, ATP levels are likely high enough to activate P2X7, but the downregulation of P2X7 expression in Tregs (only about 40% of retinal Tregs expressed P2X7, Fig. 1E) could make them less susceptible to P2X7-mediated apoptosis.

Importantly, our work highlights a major role of P2X7-expressing MPs, both MdMs and microglia, in the development of EAU. We showed that lack of P2X7 in MPs generally decreased the severity of the clinical signs of EAU and diminished immune cell infiltration in the retina. Transcriptomic analysis of MPs in the EAU retina revealed that the lack of P2X7 in this cell type down-regulated genes belonging to the phagocytosis and inflammatory pathways. While phagocytosis serves as an essential mechanism for retinal homeostasis under physiological conditions, MP activation can transform this process into a pathological one, resulting in aberrant engulfment of viable photoreceptors and subsequent retinal degeneration [57]. Interestingly single-cell analyses have revealed a monocyte subpopulation in Behçet’s disease patients exhibiting enhanced phagocytosis, suggesting potential immune dysregulation in uveitis pathogenesis [58].

Non-infectious uveitis is characterized by elevated levels of pro-inflammatory mediators, particularly CD4⁺ T cell-derived cytokines in the aqueous humor [23, 59]. Among these, TNFα appears critically involved, as evidenced by its increased levels in Behçet’s disease and idiopathic uveitis, and the clinical efficacy of anti-TNF-α therapies [23, 59, 60]. Our findings reveal that P2X7-deficient MPs exhibit downregulation of ADAM10/17 metalloproteases, key enzymes mediating ectodomain shedding of TNF-α and other membrane proteins [5, 9, 42, 61, 62]. Thus, the reduced expression of these proteases likely results in lower TNFα levels. Consequently, P2X7 receptor activation could exacerbate this inflammatory pathway.

In addition, the P2X7-dependent inflammasome pathway triggers the release of IL1β and IL18 [2]. We found that MdMs release ∼60 times more IL1β than microglia (Fig. 6D), and this release of IL1β is completely lost in P2X7-deficient MdMs, suggesting that P2X7 in MdMs plays a dominant role in this process. Our results also revealed that lack of P2X7 expression in MPs reduced the frequency of Th17 cells (Fig. 6C), which are known to contribute to EAU [45]. Given that IL1β is essential for the differentiation of Th17 cells in EAE [46, 63] and as we demonstrated in vitro (Fig. 6A), we hypothesize that this effect may stem from diminished IL1β release in P2X7-deficient MdMs. This is supported by our co-culture experiments showing that P2X7-expressing MdMs promoted significantly greater Th17 polarization than P2X7-deficient cells (Fig. 6E). Thus, in EAU, P2X7-mediated inflammasome activation in MdMs may promote Th17 development, highlighting an indirect yet critical role in disease pathogenesis.

Our results show that lack of P2X7 in MPs reduced the expression of components of the complement-system, in particular the C4b isoform and C3 (Fig. 5A), which is cleaved into C3b and the anaphylatoxin C3a in the complement cascade, with scRNA sequencing specifically identifying reduced expression of the anaphylatoxin receptors, C3aR1 and C5aR1, in P2X7-deficient MPs (Fig. 7D). These findings carry particular relevance in uveitis as C3aR1/C5aR1 deficiency confers protection against EAU [64], and elevated C3aR1 expression characterizes monocytes from active Vogt-Koyanagi-Harada uveitis patients [65]. The pathological significance of this pathway is further underscored by the observation that C3aR1 activation drive aberrant phagocytosis leading to synaptic loss in neurodegenerative and viral infection models [66, 67]. Moreover, anaphylatoxin signaling could also promote Th17 polarization through C5aR activation on MPs in autoimmune conditions [68, 69] and C3aR1 activation inducing NLRP3 inflammasome activity via ATP release in monocytes [70]. Thus, P2X7-dependent regulation of the complement pathway may influence phagocytic activity and promote Th17 cells during EAU. scRNAseq analysis revealed P2X7-dependent complement and phagocytosis gene downregulation specifically in MdMs, but not microglia, during EAU (Fig. 7D-F). These findings suggest that P2X7 expression in MdMs contributes to the pathogenic effects in EAU, via inflammasome activation, phagocytosis, and complement signaling.

In EAU microglia exhibit properties of antigen presenting cells [54] and have been shown to contribute to the recruitment of immune cells [55] but their roles are still ill-defined. To analyze P2X7-dependent transcriptional changes in microglia, we turned to scRNAseq of MPs, as microglial transcriptional changes can be lost in bulk sequencing experiments due to the limited number of microglia (∼25% of CD11b⁺ cells sequenced, suppl Fig. 7) and lower transcript levels compared to MdMs. First, we noticed that the gene expression profile of microglia shifted from a homeostatic state in naive mice to a pro-inflammatory state during EAU, including the expression of IFN-response genes (Ifitm3, Isg15, Irf7, Stat1) [48, 49] but not the DAM signature present in models of neurodegenerative diseases (Fig. 7A&G). This EAU gene expression profile may be explained by the significant infiltration of MdMs and T-cells, creating a highly inflammatory environment that impacts the microglial polarization from a regenerative to a pathogenic state. In addition, we found that the absence of P2X7 in MPs reduced the population of IFN-responsive microglia [51]. Our analysis revealed that biological processes related to the IFN-responsive microglial population in EAU include the IFN pathway, antigen processing and TNF production (Suppl Fig. 4 and Tabular data 3), as observed in the IFN-responsive microglial population in the spinal cord of young mice post-injury [52]. Although our data could not clearly establish whether P2X7 drives the reduction of IFN-responsive microglia directly or indirectly, they suggested that this population of microglia tends to be especially pro-inflammatory and contributes to the development of EAU. Notably, microglia are not the only IFN-responsive cells involved in uveitis. A subset of IFN-responsive neutrophils has recently been identified as playing a key modulatory role in Behçet’s disease [71, 72]. Although P2X7 is not expressed on neutrophils in the retina during EAU, it is possible that the decrease in IFN-responsive microglia may nonetheless impact this population. Microglia are implicated in various degenerative and inflammatory eye diseases, where they often exacerbate disease progression. To specifically investigate the pathogenic role of P2X7 expression in microglia, we used Cx3cr1^creERP2X7^lox mice and found that the deletion of P2X7 in microglia decreased the severity of the disease (Fig. 8), indicating that activation of P2X7 in microglia promotes their deleterious functions.

Together, our findings demonstrate that P2X7 expression in both microglial cells and MdMs promotes auto-immune uveitis. The loss of P2X7 expression in MPs may reduce TNFα production, impair phagocytic activity (MdM), and attenuate IL1β release (MdM) - a cytokine known to be critical for the polarization of pathogenic Th17 cells. Importantly, these pathways have been implicated in both autoimmune and autoinflammatory processes, which are key contributors to the pathogenesis of non-infectious uveitis [20]. Given the reported overexpression of P2X7 in monocytes from patients with Behçet’s disease [26], our data support the potential of P2X7 inhibition as a therapeutic strategy capable of targeting multiple pathogenic mechanisms. Furthermore, the protective effects observed following microglia-specific deletion of P2X7 suggest that local intraocular inhibition may offer therapeutic benefits while avoiding potential adverse effects associated with systemic blockade such as impaired host defense and immune surveillance.

In conclusion, our comprehensive cell-specific analysis of P2X7 functions in EAU has revealed a pivotal role for P2X7 expression in MPs, orchestrating both innate and adaptive immune responses in autoimmune neuroinflammation.

Authors’ contributions

CD conceived the project. JK, SA, XG, FS and CD secured funding. PADM and CD designed the experiments. PADM, YC, CR, KR, SA, PLR, CN carried out experiments. PADM, YC, CR, FB, PLG, XG and CD analyzed data. PADM and CD wrote the manuscript. KR, ST, BB, XG and FS edited the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by grants from INSERM, Sorbonne University, Union Nationale des Aveugles et Deficients Visuels (UNADEV), ANR PAON (ANR-23-CE14-0046-01), Programme Investissements d’Avenir IHU FOReSIGHT (ANR-18-IAHU-0001). The work described in this article was supported by the COST Action CA21130 “P2X receptors as a therapeutic opportunity (PRESTO)”.

Data availability

Data is provided within supplementary information files.

Declarations

Ethics approval and consent to participate

For animal study: all experimental protocols and procedures were approved by the French Ministry of higher Education, Research and Innovation (APAFIS #32691-2021080310403747 v5) and followed the European directive 2010/63/UE, in agreement with the ARRIVE guidelines.

Competing interests

The authors declare no competing interests.