Extracellular ATP (eATP) is a signaling molecule that variably affects all cells of the immune system either directly or after hydrolysis to adenosine. Although eATP is virtually absent in the interstitium of normal tissues, it can be present in the hundreds of micromolar range in tumors, a concentration compatible with activation of the ATP-gated ionotropic P2X7 receptor. Here, we show that P2X7 activity in tumor-infiltrating lymphocytes (TIL) induces cellular senescence and limits tumor suppression. P2X7 stimulation affected cell cycling of effector T cells and resulted in generation of mitochondrial reactive oxygen species and p38 MAPK-dependent upregulation of cyclin-dependent kinase inhibitor 1A (Cdkn1a, encoding for p21Waf1/Cip1). Lack of P2X7 promoted a transcriptional signature that correlated with enhanced cytotoxic T-cell response in human solid tumors. In mice, transfer of tumor-specific T cells with deletion of P2rx7 significantly reduced tumor growth and extended survival. Collectively, these findings uncover a purinergic checkpoint that can be targeted to improve the efficacy of cancer immunotherapy strategies.
These findings suggest that the purinergic checkpoint P2X7 may be targeted to enhance T-cell–mediated cancer immunotherapy and improve T effector cell accumulation in the tumor microenvironment.
More than 30 years have passed since the first tumor-infiltrating lymphocytes (TIL) were conditioned for treating patients with metastatic melanoma (1). Today, checkpoint inhibitors and chimeric antigen receptor (CAR) T cells have succeeded in promoting effective antitumor cytotoxicity with unprecedented durable responses in a variety of cancers, thereby establishing new immunotherapeutic paradigms for oncologic patients (2, 3). Unfortunately, a substantial fraction of patients do not respond to current immunotherapy treatments. Thus, it has become important to identify factors that limit efficient T-cell responsiveness in the tumor microenvironment (TME), and to develop strategies that could potentially increase the patient response rates.
The TME can impair the effector functions of TILs by various mechanisms, including nutrients depletion (4, 5) and release of immunosuppressive molecules, such as indoleamine 2,3-dioxygenase (6), by cancer cells; recruitment of myeloid suppressor cells that release immunomodulators, such as arginase and nitrous oxide synthase (7); and hypoxia (8) and release of intracellular potassium ions by tumor-associated necrosis, both of which suppress T-cell effector function (9). A characteristic feature of the tumor interstitium is the elevated concentration of extracellular ATP (eATP; ref. 10), a pleiotropic signaling molecule, which can act as a danger-associated molecular pattern. In fact, eATP contributes to adjuvant's efficacy in vaccination (11) and promotes immunogenic cell death of cancer cells by attracting antigen-presenting cells and activating proinflammatory cascades (12); however, it can also limit proinflammatory T-cell effector function (13) or generate immunosuppressive adenosine through the activity of plasma membrane ectonucleotidases (14). Therefore, the final effect of eATP in the TME would depend on the nature of the immune cell infiltrate, the composition of receptors for extracellular nucleotides, and the activity of ATP-degrading ectonucleotidases.
Plasma membrane receptors for extracellular nucleotides, termed P2 receptors, are divided into two families, P2X and P2Y (15). P2X1 to P2X7 receptors are ATP-gated nonselective cation channels, whereas P2Y receptors are guanine nucleotide–binding protein–coupled receptors (GPCR), which bind also to ADP, UDP, UTP, or UDP-glucose. The P2X7 receptor is widely expressed with highest levels in the nervous and immune systems; it is characterized by dual gating that depends on the saturation level of the ligand binding sites. Activation of P2X7 by low agonist concentrations results in slow desensitizing currents, whereas saturating or repetitive stimulations generate high-amplitude currents that lead to dilation of a pore permeable to nanometer-size dyes and eventually cell death (16, 17). Albeit a detailed analysis of αβTCR repertoire was not performed in naïve T cells from P2rx7−/− mice, P2rx7 expression did not apparently influence αβ T-cell development in the thymus (18). We found that P2rx7 is robustly upregulated in T effector memory (TEM) cells compared with the naïve counterpart and its activity in TILs promotes cell-cycle arrest, cellular senescence, and impairs the tumoricidal response. Our results unravel P2X7 as a possible target to foster the adaptive T-cell response against cancer cells in immunotherapeutic approaches.
Materials and Methods
Mice and in vivo experiments
All animal experiments were performed in accordance with the Swiss Federal Veterinary Office guidelines and approved by the Ethical Committee of the Cantonal Veterinary with authorization number TI 37/2016. C57BL/6J, P2rx7−/− (B6.129P2-P2rx7tm1Gab/J), Cd3ϵ−/−, OT-II Rag1−/− [B6.Cg-Tg(TcraTcrb)425Cbn/J], OT-I Rag1−/− [B6.Cg-Tg(TcraTcrb)1100Mjb/DcrJ], and CD45.1 (B6.SJL-Ptprca Pepcb/BoyJ) mice were bred in specific pathogen-free facility at the Institute for Research in Biomedicine (Bellinzona, Switzerland). OT-II Rag1−/− P2rx7−/− and OT-I Rag1−/− P2rx7−/− were generated by crossing OT-II Rag1−/− or OT-I Rag1−/− with P2rx7−/− mice. Genotyping was accomplished by the PCR method according to the manufacturer's protocol. Mice were housed, 5 per cage, in ventilated cages under standardized conditions (20°C ± 2°C, 55% ± 8% relative humidity, and 12-hour light/dark cycle). Food and water were available ad libitum, and mice were examined daily. To analyze antitumor response, CD4+ or CD8+ T naïve cells were sorted from C57BL/6J and P2rx7−/− mice as described below. A total of 2.5 × 105 T naïve cells were injected into Cd3ϵ−/− mice. Melanoma B16F10 cells were harvested at exponential growth. After 5 days from transfer of T cells, melanoma cells were resuspended in PBS at a concentration of 5 × 106 cells/mL and a volume of 0.1 mL (5 × 105 tumor cells) was injected subcutaneously into the back of adoptively transferred Cd3ϵ−/− mice. To analyze the T-cell response against a tumor-specific antigen, C57BL/6 or CD45.1 mice were injected subcutaneously with 5 × 105 B16-OVA or MC38-OVA cells. After 5 or 7 days, 1 × 106 Rag1−/− or Rag1−/−/P2rx7−/− congenically marked CD4+ OT-II or CD8+ OT-I, which had been previously activated in vitro with anti-CD3 and anti-CD28 antibodies for 72 hours and IL2 in the last 24 hours prior to injection (19), were transferred into mice randomized from littermate cages and tumor growth was assessed. Tumor growth was scored with a caliper by measuring the greatest tumor diameter and its perpendicular to determine an average and then the area was calculated as: (average/2)2π. Tumor-bearing animals were sacrificed after 20 days or earlier when showing any sign of discomfort. For analysis of survival, mice were sacrificed when they reached humane endpoint, defined by: tumor volume [estimated with the formula (length × width2)/2] more than 1.5 cm3 or severe signs of discomfort.
Cell isolation from mice organs
For in vitro experiments and adoptive transfer, CD4+ or CD8+ T naïve, TEM, or congenically marked OT-II or OT-I cells were sorted at FACSAria (BD Biosciences) from pooled cell suspensions of spleen, inguinal, axillary, brachial, cervical, and mesenteric lymph nodes collected from C57BL/6J and P2rx7−/− mice. T naïve cells were sorted as CD4+ or CD8+, CD62L+, CD44−, and CD25− cells, TEM cells were sorted as CD4+ or CD8+, CD62L−, CD44+, and CD25− cells. OT-I and OT-II cells were sorted as CD8+ and CD4+, respectively, CD25− and CD45.1+ or CD45.1− cells. Magnetic Cell Sorting (Miltenyi Biotec) with anti-CD4 and anti-APC mAbs was used to enrich cell subsets from complex cell mixtures.
Tumor cell lines
B16F10, B16F10-OVA (B16-OVA), and MC38-OVA cells were cultured in RPMI1640 supplemented with 10% heat-inactivate FBS, 100 U/mL penicillin/streptomycin, and 100 U/mL kanamycin. Cells were tested for the absence of Mycoplasma and maintained in 5% CO2 at 37°C. Frozen aliquots were thawed for each in vivo experiment and passaged in vitro for the minimum time required. Tumor cells at 70%–80% confluency were harvested by diluting them 1:5 in 0.25% trypsin. B16F10 and B16-OVA cells were obtained from Dr. Matteo Bellone (HSR Scientific Institute, Milan, Italy). MC38-OVA cells were obtained from Dr. Maria Rescigno (Humanitas University, Rozzano, Milan, Italy).
Generation of stable B16F10-pmeLUC transfectants
B16F10 melanoma cells were transfected with 1 μg/mL of pmeLUC-pcDNA3 plasmid by Lipofectamine LTX (Thermo Fisher Scientific) according to the manufacturer's instructions. For stable transfection, cells were kept in continuous presence of G418 (0.4 mg/mL; Calbiochem) for 2 weeks. B16F10-pmeLUC clones were obtained by limiting dilution and B16F10pmeLUC-positive clones were selected by luminescence assay in the presence of the Luciferase substrate d-luciferin (Promega). Stably transfected clones were maintained in culture in the presence of G418 (0.2 mg/mL; ref. 20).
Luciferase and in vivo imaging
Luciferase luminescence was acquired with a total body luminometer (IVIS Lumina, Caliper-PerkinElmer). Mice were anesthetized with 2.5% isoflurane, intraperitoneally injected with 150 mg/kg d-luciferin (PerkinElmer), and luminescence was quantified after 15 minutes. Regions of interest from displayed images were identified at tumor sites and quantified as total flux (photons/second) using the living image software.
In vitro calibration of B16F10-pmeLUC cells
B16F10-pmeLUC cells, 1 × 105 per well, plated in a 24-well plate, were incubated in RPMI1640 medium and challenged with increasing concentration of ATP in the presence of d-luciferin, 60 μg/mL. Luminescence was acquired with the IVIS luminometer for 1 minutes. Total luminescence emission was acquired from each well and quantified as total flux (photons/second) using the living image software. Luminescence was expressed as total flux (photons/second) as a function of the added ATP concentration.
Cell isolation from tumor tissue
Tumors were cut in small pieces and resuspended in RPMI1640 with 1.5 mg/mL Type I Collagenase (Sigma), 100 μg/mL DNase I (Roche), and 5% FBS, digested for 45 minutes at 37°C under gentle agitation. The digestion product was then passed through a 70 μm cell strainer to obtain a single-cell suspension. Lymphocytes were then enriched by Percoll density gradient following the manufacturer's protocol.
Time monitoring of DAPI uptake
Purified T naïve or TEM cells were resuspended at 1 × 106 cells/mL and loaded with DAPI (1 μg/mL). After 30 seconds of measurement for DAPI basal level, cells were stimulated with 3′-O-(4-benzoyl) benzoyl ATP (BzATP) and DAPI uptake was monitored over time (250 seconds) at LSRFortessa and the kinetics were analyzed using FlowJo software.
Gene expression profiling of ex vivo purified CD4+ cells
Gene expression profiling of ex vivo purified CD4+ cells was performed on MG-430 PM Array Strip (Affymetrix). Briefly, total RNA was extracted using TriPure (Roche) from ex vivo purified CD4+ naïve and TEM cells from wild-type (WT; n = 2 and n = 3 independent samples for naïve and TEM, respectively) and P2rx7−/− (n = 2 and n = 3 independent samples for naïve and TEM, respectively) mice. RNA quality and purity were assessed on the Agilent Bioanalyzer 2100 (Agilent Technologies); RNA concentration was determined using the NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies Inc.). In vitro transcription, hybridization, and biotin labeling were performed according to Affymetrix GeneChip 3′IVT protocol and processed using GeneAtlas Platform (Affymetrix). Scanning and data exporting were performed using standard Affymetrix protocols.
All microarray data analyses were performed in R (version 3.3.2) using Bioconductor libraries (BioC 3.1) and R statistical packages. Probe level signals were converted to expression values using robust multi-array average procedure RMA (21) of Bioconductor affy package and a custom definition file for mouse HT array plates based on Entrez genes from BrainArray (version 22.0.0; http://brainarray.mbni.med.umich.edu/Brainarray/Database/CustomCDF/22.0.0/entrezg.asp). Before downstream analysis, expression values were batch-corrected using the ComBat function of the sva package. Raw data are available at Gene Expression Omnibus (GEO) under accession number GSE118146. To identify genes associated with P2rx7 deletion in ex vivo purified CD4+ TEM cells, we compared the expression levels of WT and P2rx7−/− TEM cells using the significance analysis of microarray (SAM; ref. 22) algorithm coded in the samr R package. In SAM, we estimated the percentage of false positive predictions (i.e., FDR) with 100 permutations and selected those probe sets with FDR ≤ 5% and absolute fold change larger than a selected threshold (e.g., ≥ 1.5) in the comparison of TEM cells from P2rx7−/− and WT mice (Supplementary Tables S1 and S3). Principal component analysis (PCA) was performed using the function prcomp of R stats package. Before unsupervised PCA, to reduce the effect of noise from nonvarying genes, we removed those probe sets with a coefficient of variation smaller than the 95th percentile of the coefficients of variation in the entire dataset. The filter retained 904 genes that were more variable across samples in any of the four subsets (naïve and TEM cells from P2rx7−/− and WT mice). Supervised clustering was performed using the function hclust of R stats package with Pearson correlation as distance metric and average agglomeration method. Gene expression heatmap was generated using the function heatmap.2 of R gplots package after row-wise standardization of the expression values. The volcano plot, showing the most significantly differentially expressed genes in the comparison of ex vivo purified CD4+ TEM cells from P2rx7−/− and WT mice, was generated using the ggplot function of the ggplot2 R package. P values were derived from SAM q values using the function samr.pvalues.from.perms of the samr R package (Supplementary Table S3). Functional overrepresentation was performed using gene set enrichment analysis (GSEA; http://software.broadinstitute.org/gsea/index.jsp) and the curated gene sets of the Molecular Signatures Database (MSigDB) derived from the Kyoto Encyclopedia of Genes and Genomes pathway database (http://software.broadinstitute.org/gsea/msigdb/genesets.jsp?collection=CP:KEGG). GSEA was applied on log2 expression data of P2rx7−/− and WT TEM cells. Prior to GSEA, we converted mouse Entrez IDs into the corresponding human homologous genes using the HUGO Gene Nomenclature Committee database (https://www.genenames.org/cgi-bin/hcop). Gene sets were considered significantly enriched at FDR ≤ 0.25 when using Signal2Noise as metric and 1,000 permutations of gene sets.
Gene expression data that support the findings of our study have been deposited in NCBI GEO and are accessible through the link: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE118146 by using the following secure token: oxmpkqkotfidbuz.
Statistical analysis was performed with the Prism Software (GraphPad). Comparisons of two groups were calculated using nonparametric Mann–Whitney test or Student unpaired t test. Comparisons for more than two groups were calculated using Kruskal–Wallis test followed by Dunn multiple comparison test. Results are presented as mean values ± SEM or SD. Values of P < 0.05 were considered significant.
Enhanced antitumor response by P2rx7−/− CD8 T cells in lymphopenic mice
P2rx7 is the most expressed P2rx gene in both CD4+ T naïve and TEM cells (Supplementary Fig. S1A). Nevertheless, real-time quantitative reverse-transcription PCR (qRT-PCR) and Western blot analysis revealed significantly increased levels of P2rx7 transcript and P2X7 protein, respectively, in TEM versus naïve cells (Supplementary Fig. S1B and S1C). The analysis of pore opening in flow cytometry in CD4+ T cells upon stimulation with the selective agonist BzATP at 150 μmol/L revealed that only a small fraction of naïve cells was sensitive to P2X7-mediated DAPI uptake, as compared with the vast majority of TEM cells (Supplementary Fig. S1D), indicating enhanced sensitivity of CD4 cells to eATP after stimulation by cognate antigen and differentiation to effector/memory cells. An analogous upregulation of P2rx7 transcripts was observed in CD8+ TEM cells (Supplementary Fig. S1E).
In orthotopic human melanoma xenografts, eATP is present in the hundreds micromolar range (10, 23). The analysis of B16F10 melanoma composed by cells stably expressing the eATP reporter, pmeLUC (23), at day 7 after transplant into WT mice revealed concentrations of eATP in the TME from tens to hundreds of micromolar (Supplementary Fig. S1F). We addressed whether P2X7 stimulation in the TME affected control of tumor growth by CD8+ TILs. The transfer of P2rx7−/− CD8+ naïve T cells into Cd3ϵ−/− mice that were subsequently engrafted with OVA expressing B16F10 (B16-OVA) melanoma cells (Fig. 1A) resulted in enhanced accumulation of P2rx7−/− TEM cells in the tumor, but not in the spleen with respect to WT cells (Fig. 1B and C). Tetramer staining revealed increased percentages of P2rx7−/− versus WT TILs specific for the H-2Kb–restricted OVA peptide 257–264 (Fig. 1D). Moreover, tumor growth was significantly delayed by P2rx7−/− with respect to WT cells (Fig. 1E).
Enhanced control of tumor growth and survival of mice by lack of P2X7 activity in tumor-specific CD8 cells
Adoptive transfer of in vitro--primed congenic Rag1−/−/P2rx7+/+ (OT-I) or Rag1−/−/P2rx7−/− (OT-I P2rx7−/−) OT-I TCR transgenic CD8+ cells, specific for chicken ovalbumin peptide 257–264 presented by H2Kb, into WT mice bearing B16-OVA tumors (Fig. 2A) resulted in increased absolute number (Supplementary Fig. S2A) and percentage of tumor-infiltrating OT-I P2rx7−/− cells (Fig. 2B), which showed enhanced secretion of IFNγ and granzyme B with respect to OT-I cells (Fig. 2C). The increase in the percentage of P2rx7−/− cells did not impact on the frequency of endogenous CD8+ TILs (Supplementary Fig. S2C).
We checked phosphatidylserine exposure and IFNγ secretion before injection; no significant differences were detected between P2rx7+/+ and P2rx7−/− cells (Supplementary Fig. S2D and S2E). Moreover, because P2X7 activity can result in diminished recovery of particular T-cell subsets during the ex vivo isolation procedure (24), we purified TILs in the presence of the selective P2X7 antagonist, A-438079. We did not detect any difference in the percentages of OT-I or OT-I P2rx7−/− cells within CD8+ TILs irrespective of drug's addition (Supplementary Fig. S2F). In spite of the function of P2X7 in promoting caspase-mediated cell death (13), we found no difference in caspase activation between P2rx7+/+ and P2rx7−/− TILs (Fig. 2D). However, P2rx7−/− cells were characterized by an increased proliferative potential than WT cells, as shown by staining with Ki-67 antibodies (Fig. 2E). Tumor infiltration by OT-I P2rx7−/− cells resulted in significant inhibition of tumor growth as compared with mice transferred with OT-I cells (Fig. 2F). Importantly, the transfer of OT-I P2rx7−/− cells promoted extended survival of mice (Fig. 2G), suggesting that P2X7 limits the expansion and tumoricidal activity of TILs. Consistent with enhanced capacity of OT-I P2rx7−/− cells to accumulate in the TME, cotransfer of OT-I and OT-I P2rx7−/− cells at 1:1 ratio (Supplementary Fig. S2G) into WT mice bearing B16-OVA tumors resulted in the significant dominance of OT-I P2rx7−/− TILs (Fig. 2H). Notably, OT-I P2rx7−/− cells were characterized by reduced expression of PD-1, which distinguishes dysfunctional CD8 cells (Fig. 2I), whereas other checkpoint proteins, including Tim3 and CTLA4, were not differentially expressed by coinfiltrating P2rx7+/+ and P2rx7−/− cells (Supplementary Fig. S2H). These data suggest that lack of P2X7 endows TILs with improved resistance to dysfunctionality induced by the TME.
To substantiate these results within a different TME, we engrafted WT mice with OVA expressing MC-38 (MC38-OVA) colon adenocarcinoma cells and adoptively transferred OT-I or OT-I P2rx7−/− CD8+ cells (Supplementary Fig. S3A). As observed in B16-OVA tumors, P2rx7−/− TILs were significantly increased with respect to P2rx7+/+ cells (Supplementary Figs. S2B and S3B) and expressed lower levels of PD-1 in the plasma membrane (Supplementary Fig. S3C). Moreover, mice transferred with P2rx7−/− cells showed significantly reduced tumor growth and enhanced survival (Supplementary Fig. S3D and S3E), suggesting P2X7-deficient OT-I cells displayed enhanced antitumor functionality also in the MC-38 colon adenocarcinoma–conditioned microenvironment.
Enhanced expansion of P2rx7−/− CD4 TEM cells in the TME
To address whether P2X7 activity could limit CD4+ T-cell abundance in the TME, we adoptively transferred CD4+ naïve cells from WT or P2rx7−/− mice into Cd3ϵ−/− mice that were subsequently engrafted with B16F10 melanoma cells (Supplementary Fig. S4A). Flow cytometry at day 20 after tumor transplant revealed a significant increase in P2rx7−/− with respect to WT TILs with TEM phenotype, but not in the spleen (Supplementary Fig. S4B and S4C). We did not detect differences in tumor growth between Cd3ϵ−/− mice either nontransferred or transferred with P2rx7−/− or WT CD4+ naïve cells, consistent with lack of CD8+ T-cell–mediated cytotoxicity in Cd3ϵ−/− mice (Supplementary Fig. S4D).
To mimic the effector response of CD4 cells to a tumor-specific antigen in an immunocompetent organism, we adoptively transferred either Rag1−/−/P2rx7+/+ (OT-II) or Rag1−/−/P2rx7−/− (OT-II P2rx7−/−) congenic CD4+ OT-II TCR transgenic T cells specific for I-Ab–restricted OVA peptide 323–339 into WT mice bearing B16-OVA tumors (Fig. 3A). As observed in mice adoptively transferred with OT-I P2rx7−/− cells, the increased proportion of OT-II P2rx7−/− cells did not influence the frequency of endogenous CD4+ TILs (Supplementary Fig. S5A). As for in vitro–primed CD8+ OT-I naïve cells, we checked phosphatidylserine exposure, and IFNγ as well as IL17 secretion before injection; no significant differences were detected between WT and P2rx7−/− cells (Supplementary Fig. S5B and S5C). In mouse naïve CD4 T cells, ADP ribosylation of P2X7 by the ectoenzyme ADP-ribosyltransferase 2.2 (ARTC2.2) is responsible for NAD-induced T-cell death (24). The analysis of OT-II T cells after in vitro activation and before injection into tumor-bearing mice showed lack of ARTC2.2 expression, ruling out a possible function of ADP ribosylation in impairing the expansion of WT with respect to P2rx7−/− cells (Supplementary Fig. S5D). After 15 days, both OT-II and OT-II P2rx7−/− cells were barely detectable in the spleen (Supplementary Fig. S5E), however, OT-II P2rx7−/− cells were significantly increased in tumors as compared with OT-II cells (Fig. 3B). The percentage of Foxp3+ immunosuppressive T regulatory cells (Treg) infiltrating the tumor tissue was not influenced by the expression of P2X7 in transferred Rag1−/− OT-II cells (Supplementary Fig. S5F). The analysis of endogenous CD8+ TILs showed the significant increase of IFNγ and TNFα secreting cells in mice transferred with OT-II P2rx7−/− cells (Fig. 3C), suggesting lack of P2X7 fostered helper function to cytotoxic T cells, which could contrast tumor growth more efficiently (25). Accordingly, tumor growth was significantly delayed in mice transferred with P2X7-deficient cells (Fig. 3D). Analogously to the results obtained with CD8 cells, cotransfer of OT-II and OT-II P2rx7−/− cells at 1:1 ratio (Supplementary Fig. S5G) into WT mice bearing B16-OVA tumors resulted in significant dominance of P2rx7−/− over P2rx7+/+ cells (Fig. 3E), indicating that P2X7-deficient CD4 TEM cells are endowed with enhanced tumor infiltrating potential with respect to P2X7-proficient cells.
Regulated cell cycling in TEM cells by P2X7 activity
In CD4 naïve T cells, P2X receptors' activation concomitantly to T-cell receptor (TCR) stimulation contributes to productive T-cell activation. In fact, P2X inhibition by the pharmacologic antagonist, periodate-oxidized ATP, in cells stimulated with anti-CD3/CD28 antibodies promotes T-cell anergy (26). P2rx7−/− CD4 naïve cells did not show any difference in cell proliferation as compared with WT cells (Fig. 4A), suggesting that P2X1 and/or P2X4 could compensate for the lack of P2X7 activity, as observed in human T cells (27). In contrast, stimulation of P2rx7−/− TEM cells revealed a peculiar enhancement of cell proliferation with respect to the WT counterpart (Fig. 4B). To better address the contribution of P2X7 in regulating cell survival versus cycling, we applied a mathematical model to quantify division times and death rates within a time-course experiment in CD3/CD28-stimulated TEM cells (28). Graphical data extrapolation showed that P2rx7−/− TEM cells progressed earlier in the first cell division and required less time to enter subsequent cell divisions (Fig. 4C). At the same time, P2rx7−/− TEM cells were characterized by a reduced rate of cell death overtime as indicated by the slower exponential decay of the relative constant (Fig. 4D). These results indicate that P2rx7−/− TEM cells “perform” better than WT cells following TCR stimulation. To see whether an analogous difference could be detected following cytokine-driven stimulation (without TCR engagement), we stimulated purified TEM cells with IL2 or IL7. In contrast to WT cells, which poorly proliferated, P2rx7−/− TEM cells effectively expanded with both IL2 and IL7 (Fig. 4E and F). Altogether, these data suggest that P2X7 activity limits TEM cell proliferation.
Transcriptional regulation of cell cycling by P2X7 in TEM, but not naïve CD4 cell
To explore the transcriptional impact of P2rx7 deletion in CD4+ T naïve and TEM cells, we performed genome-wide expression profiling to compare ex vivo purified cells from WT and P2rx7−/− mice. Unsupervised PCA of gene expression levels showed that naïve cells grouped independently from the genotype, while P2rx7−/− TEM cells were clearly separated from the WT counterpart, suggesting that P2rx7 deletion substantially influences gene transcription in TEM, but not naïve cells (Supplementary Fig. S6A). Differential expression analysis resulted a transcriptional signature of 158 upregulated and 255 downregulated genes in P2rx7−/− TEM with respect to WT cells (FDR ≤ 5% and absolute fold change ≥ 1.5; Supplementary Table S1) that discloses how P2rx7 deficiency induces in TEM cells a transcriptional pattern intermediate between naïve and WT TEM cells (Supplementary Fig. S6B). Functional overrepresentation analysis revealed that gene sets associated to DNA replication were enriched in P2rx7−/− TEM cells (Supplementary Fig. S6C), whereas TCR/cytokine signaling, apoptosis, and cell-cycle arrest signatures were enriched in WT TEM cells (Supplementary Fig. S6D). These results suggest that lack of P2X7 might confer greater proliferation potential to TEM cells in the eATP-rich TME. The efficacy of cytotoxic T-cell response in human solid tumors has been associated to the magnitude of infiltrating CD8+ cells expressing the integrin alpha E chain (CD103) together with the ATP-hydrolyzing plasma membrane ectonucleoside triphosphate diphosphohydrolase-1 (CD39), suggesting that limiting P2X7 signaling in TILs could enhance the tumoricidal activity of these cells (29–31). Notably, genes overexpressed in P2rx7−/− TEM cells (Supplementary Table S2) were significantly enriched in purified CD103highCD39+ TILs (Supplementary Fig. S6E) as well as in non–small cell lung cancers and skin cutaneous melanoma transcriptomes from patients showing significantly improved survival (Supplementary Fig. S6F and S6G). These data suggest that diminished P2X7 activity could promote a transcriptional program, which more effectively controls tumor progression in humans.
Induction of T-cell senescence by P2X7 activation
The genome-wide transcriptional analysis evidenced Cdkn1a, encoding for p21Waf1/Cip1, as one of the most downregulated transcript in P2rx7−/− TEM cells (Fig. 5A; Supplementary Table S3). P2rx7 expression at different times after in vitro stimulation of naïve CD4+ T cells directly correlated with Cdkn1a, suggesting signaling by P2X7 positively regulated Cdkn1a transcription (Fig. 5B). Accordingly, qRT-PCR on purified TEM cells stimulated with BzATP alone or together with A-438079 confirmed that signaling by P2X7 induced Cdkn1a; this induction was selective and did not affect Cdkn1b transcription (Fig. 5C). Coherent with gene array data and the function of P2X7 in regulating cell-cycling activity, the transcription of Trp53 and Gadd45b genes was also induced by BzATP (Fig. 5D). Moreover, addition of BzATP to in vitro stimulated WT CD4+ TEM cells with anti-CD3/CD28 antibodies resulted in reduced Ki-67 staining and CellTrace Violet dilution, indicating inhibition of cell-cycling activity. This effect was abrogated by A-438079 (Fig. 5E). BzATP showed a dose-dependent effect on cell proliferation that was also observed in CD8+ WT TEM cells, albeit with higher doses of BzATP (Supplementary Fig. S7A and S7B), and Cdkn1a was upregulated in vitro by BzATP (Supplementary Fig. S7C). Notably, Cdkn1a transcript levels were significantly increased in WT as compared with P2rx7−/− CD8+ TILs (Fig. 5F).
Cdkn1a is a well-characterized inhibitor of cell-cycle progression and its activation by stress-induced p53 contributes to the onset of cellular senescence (32). To address whether P2X7 activity promoted TEM cells senescence, we stimulated ex vivo purified WT and P2rx7−/− TEM cells with anti-CD3/28 antibodies for 72 hours and analyzed senescence-associated β-galactosidase (SA-β-gal). We observed reduced basal levels in P2rx7−/− cells, and addition of BzATP to the culture of WT cells resulted in increase in SA-β-gal, suggesting P2X7 stimulation in TEM cells by eATP induces cellular senescence (Fig. 6A). Consistent with occurrence of the same phenomenon by P2X7 stimulation in effector T cells infiltrating the TME, SA-β-gal+ cells were significantly increased among OT-I with respect to OT-I P2rx7−/− TILs (Fig. 6B). P2X7 activity in T cells was associated with enhanced generation of mitochondrial reactive oxygen species (ROS; ref. 33). MitoSOX staining of CD4 TEM cells showed an increase in mitochondrial ROS generation following BzATP stimulation and the frequency of P2rx7−/− cells stained with MitoSOX Red was significantly diminished with respect to WT cells, suggesting that lack of P2X7 resulted in reduced production of superoxide by mitochondria (Fig. 6C). ROS promote the formation of DNA damage foci that contain H2A.X histone phosphorylated at Ser139 (γH2A.X; ref. 34). Consistent with enhanced ROS production, WT TEM cells showed increased γH2A.X histone by P2X7 stimulation in Western blot analysis that was inhibited by A-438079 (Fig. 6D). Cellular senescence in T cells can be actively maintained by p38 MAPK signaling (35). Stimulation of P2X7 induced p38 MAPK Thr180/Tyr182 phosphorylation (Fig. 6E), while pharmacologic inhibition of p38 MAPK resulted in restoring basal levels of Cdkn1a transcripts (Fig. 6F). These results suggest that in the TME, P2X7 activation in TILs promoted cell-cycle arrest and p38 MAPK mediated cellular senescence (Supplementary Fig. S8).
Regulated cell cycling in human TEM cells by P2X7 activity
In humans, P2RX7 gene polymorphisms generate functionally different P2X7 isoforms. To address possible differences in the regulation of cell cycling by P2X7 activity in human TEM cells, we genotyped a large survey of the general population of the northern area of Milan (n = 2,606, see Supplementary Materials and Methods; refs. 36, 37) for two variants in the P2RX7 locus: the rs11065464, g.36458C>A intron variant, associated with loss-of-function P2X7 pore formation (38) and the rs1718119 Ala348Thr missense variant, determining increase in receptor activity (39). Analogously to murine P2rx7−/− cells, human TEM cells bearing the hypomorphic P2X7 variant rs11065464 showed a significant increase in the frequency of proliferating cells after TCR stimulation compared with cells purified from subjects carrying the hyperactive rs1718119 SNP variant (Fig. 7A). The impaired progression in the cell cycle of cells bearing hyperactive P2X7 correlated with increased levels of CDKN1A transcripts (Fig. 7A). Knockdown of P2RX7 in human CD4 or CD8 TEM cells (Supplementary Fig. S9A) resulted in the increase of cycling cells after TCR stimulation with concomitant reduction of CDKN1A expression (Fig. 7B and D). The function of P2X7 in conditioning human TEM cells' proliferation was confirmed by addition of BzATP during TCR stimulation that blocked cell-cycle progression and upregulated CDKN1A (Fig. 7C and E; Supplementary Fig. S9B). These results indicate that P2X7 activity limits the expansion potential of human TEM cells.
T cells are potent effectors in controlling tumor growth; in fact, the extent of tumor infiltration by T cells has been generally considered a good prognostic marker in a number of tumor types (40). However, the peculiar nature of the TME results in a T-cell response that is not proficient in controlling tumor growth. In patients with melanoma, early effector T cells progress to a highly proliferating dysfunctional state (41). The composition of distinct dysfunctional CD8+ subsets within TILs differentially influences the control of tumor growth and sensitivity to checkpoint blockade (42). Antigen recognition within the TME is hypothesized to be important in driving the expansion of dysfunctional cells and in fact, elegant experiments in different tumors have shown that cytotoxicity is confined to nontumor-specific bystander cells that infiltrate the TME (43, 44). In some cases, bystander cells dominated the pool of infiltrating CD8 cells, suggesting that the tumoricidal response would benefit from improving the quality of tumor-specific T cells (43). T cells transduced with CARs can mediate specific destruction of hematologic malignancies and yield durable therapeutic responses (45). However, in solid tumors, the induction of a dysfunctional state and loss of T-cell effector function by chronic antigen stimulation limit the efficacy of this immunotherapeutic approach (46–48).
The three nuclear receptor transcription factors 4A (NR4A), together with nuclear factor of activated T cells (NFAT), were shown to play an important role in controlling the cell-intrinsic program of CD8+ TILs hyporesponsiveness. The NFAT–NR4A axis controls the expression of multiple inhibitory receptors; CAR T cells with deletion of the three Nr4a genes promote regression of solid tumors and prolong survival, thereby suggesting NR4A inhibition could constitute a promising strategy in cancer immunotherapy (49). Interestingly, Nr4a2 and Nr4a3 were downregulated in P2rx7−/− TEM cells, suggesting a possible contribution of P2X7 activity to this negative regulatory pathway in TILs. The P2X7 receptor plays pleiotropic roles in shaping T-cell function. In Tfh cells, it triggers caspase activation and pyroptosis, a mechanism that limits the expansion of pathogenic CD4 cells in systemic lupus erythematosus (13). An analogous signaling is likely responsible for cell death induced by bacteria-derived ATP in Tfh cells in the Peyer patches of the small intestine to ensure controlled generation of T-cell–dependent secretory IgA and host–microbiota mutualism (50). Among downregulated genes between WT and P2rx7−/− TEM cells, we identified erythroid differentiation regulator 1 (Erdr1), which encodes for a secreted protein that induces Fas-dependent T-cell apoptosis, suggesting P2X7 activity can promote T-cell death also by this signaling pathway (51). Importantly, acute TCR stimulation of Tfh cells robustly downregulates P2rx7 expression, thus protecting antigen-responding T cell from cell death (50). Similar results have been obtained in tissue-resident memory T cells, suggesting that selective downregulation of P2rx7 in T cells that productively respond to cognate antigen would ensure the amplification of pathogen-destructing cells during infections (52). In contrast, P2X7 activity is required for the establishment and maintenance of long-lived central and tissue-resident memory CD8 T cells in mice, probably reflecting the function of P2X7 as ion channel in promoting mitochondrial function and metabolic fitness (53). Herein, we have shown that P2X7 stimulation in tumor-specific T cells within the TME results in stress-induced premature senescence (SIPS) that limits the expansion of tumoricidal cells. Differently from replicative senescence, SIPS is induced after exposure to outside factors that act as cellular stressor (54). For example, Treg cells can force both CD4+ and CD8+ effector T cells into SIPS by metabolic competition, which causes DNA damage (55). Likewise, P2X7 activity in T cells evokes a similar signaling signature characterized by MAPK-P38 activation, DNA damage induction, increased p21 expression, and β-gal activity with concomitant inhibition of T-cell proliferation. In vitro generated data by pharmacologic antagonism of P2X7 were mirrored by P2rx7−/− TILs in vivo. We hypothesize targeting of P2X7 in effector TILs might provide a rejuvenating signal able to perpetuate the tumoricidal response.
It is important to consider that stimulation of P2X7 in cells of the innate immune system and/or cancer cells by eATP can contribute to T-cell priming and control of tumor growth (12). It was recently shown that enhancement of P2X7 mediated activation of NLRP3 inflammasome in myeloid cells by an anti-CD39 antibody, which inhibits ectonucleotidase activity, promoted the antitumor response by CD8+ TILs (56). Accordingly, tumor-bearing P2X7-null mice showed lack of inflammatory infiltration and accelerated tumor progression (57). Nevertheless, an opposite outcome (i.e., enhanced control of tumor growth) with enhancement of proinflammatory TILs was observed by treating WT mice after tumor engrafting with a P2X7-selective antagonist (23). These data show how the eATP/P2X7 axis can dramatically influence tumor progression depending on the timing and cell types it activates. Our observations are important to discriminate the possible overall effect of P2X7 inhibition in the TME versus T-cell selective P2X7 inhibition that could improve the tumoricidal potential of CAR T cells or TILs expanded in vitro in immunotherapeutic approaches.
Disclosure of Potential Conflicts of Interest
A.L. Catapano reports grants and personal fees from AstraZeneca, Amgen, and Aegerion, grants from Eli-Lilly, Mediolanum, Sanofi, Rottapharm, Mylan, Menarini, Sigma Tau, Genzyme, Merck, and honoraria from Akcea, Amryt, Kowa, Novartis, and Daiichi Sankyo outside the submitted work. A. Baragetti reports grants from Cibo, Microbiota, and Salute by Vini di Batasiolo S.p.A AL_RIC19ABARA_01 and Post-Doctoral Fellowship 2020 by Fondazione Umberto Veronesi 2020-3318 outside the submitted work. F. Di Virgilio reports grants from Italian Association for Cancer Research and Italian Ministry of Education and Scientific Research (MIUR during the conduct of the study), and Ablynx/Sanofi outside the submitted work and personal fees from Biosceptre Ltd UK. F. Grassi reports a patent for “Combination of ATP hydrolysis in tumors” pending and is the founder of MV BioTherapeutics. No potential conflicts of interest were disclosed by the other authors.
A. Romagnani: Conceptualization, data curation, validation, investigation, visualization, methodology. E. Rottoli: Data curation, formal analysis, validation, investigation, visualization, methodology. E.M.C. Mazza: Data curation, formal analysis. T. Rezzonico-Jost: Investigation. B. De Ponte Conti: Data curation, formal analysis, validation, investigation, visualization, methodology. M. Proietti: Investigation, methodology. M. Perotti: Investigation. E. Civanelli: Investigation. L. Perruzza: Investigation. A.L. Catapano: Resources, funding acquisition. A. Baragetti: Resources. E. Tenedini: Formal analysis. E. Tagliafico: Formal analysis. S. Falzoni: Resources. F. Di Virgilio: Resources. G.D. Norata: Resources, formal analysis, funding acquisition. S. Bicciato: Conceptualization, data curation, software, formal analysis, funding acquisition. F. Grassi: Conceptualization, supervision, funding acquisition, writing-original draft, writing-review and editing.
We thank Sara Maffei (Institute for Research in Biomedicine, Bellinzona, Switzerland) for help with mice experiments, the NIH Tetramer Core Facility for providing tetramers. This work was supported by grant KFS–4110-02-2017-R of the Swiss Cancer Research, 310030_159491 and IZCNZ0-174704 of the Swiss National Science Foundation, Fondazione Gelu, Fondazione per la Ricerca sulla Trasfusione e sui Trapianti (to F. Grassi), AIRC Special Program Molecular Clinical Oncology ‘5 per mille’ grant 10016 and Italian Ministry of Education, University and Research, and the National Research Council grant Italian Epigenomics Flagship Project (Epigen; to S. Bicciato). The work of the PLIC study was supported by grants 2015-0524 and 2015-0564 (to A.L. Catapano) and 2016-0852 (to G.D. Norata) of Fondazione Cariplo, Italy, H2020 REPROGRAM PHC-03-2015/667837-2 (to A.L. Catapano), and GR-2011-02346974 of Ministry of Health, Italy (to G.D. Norata).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.