A central role for P2X7 receptors in human microglia

Dulbecco’s modified Eagle’s medium (DMEM), RPMI 1640, fetal bovine serum (FBS), penicillin, and streptomycin were purchased from Gibco by Life Technologies (Waltham, MA, USA). ATP, BzATP, EDTA, ethidium bromide, lipopolysaccharides from E. coli O55:B5 (LPS; L2880), dimethyl sulfoxide (DMSO), Lucifer yellow dilithium salt, 5(6)-carboxyfluorescein, carbenoxolone, probenecid, tannic acid, 4,4′-diisothiocyano-2,2′-stilbenedisulfonic acid (DIDS), and nigericin were purchased from Millipore-Sigma (St. Louis, MO, USA). BAPTA-AM, YO-PRO-1, YOYO-1, Fluo-4-AM, pHrodo Red E. coli BioParticles Conjugate, and pHrodo Green AM were purchased from Invitrogen/ThermoFisher (Carlsbad, CA, USA). A438079, A804598, BX430, and ¹⁰Panx inhibitory peptide were purchased from Tocris (Minneapolis, MN, USA). Complete growth differentiated media with serum (E37089-01-S), extracellular matrix-coated T25 or T75 culture flasks (E37089-01-T25,T75), and Xeno-free cell dissociation media (M37001-02CM) were obtained from Celprogen (Torrance, CA, USA).

Frozen ampules of healthy male (Caucasian, 29 years old) and female (Caucasian, 30 years old) human microglia isolated from the CNS (cortex) were purchased from Celprogen Inc. Freshly thawed microglia were washed once in complete growth differentiated media with serum and spun down before being maintained and sub-cultured every 48 to 72 h on human extracellular matrix-coated T25 and T75 flasks (Celprogen) at 37 °C with 5% CO₂ in a humidified atmosphere. The mouse macrophage cell line J774A.1 was obtained from ATCC (Manassas, VA, USA) and cultured in DMEM containing 10% FBS, 2 mM glutamine, 50 U/ml penicillin, and 50 μg/ml streptomycin. Human monocytic THP-1 cells from ATCC were grown in RPMI 1640 culture medium containing 10% FBS and supplemented with 0.05 mM ß-mercaptoethanol. HEK-293T cells from ATCC were maintained in DMEM containing 10% FBS, 2 mM glutamine, 50 U/ml penicillin, and 50 μg/ml streptomycin. HEK-293T cells were co-transfected with human P2X7R and fluorescent reporter plasmids using Effectene (Qiagen, Germantown, MD, USA).

Microglia were disassociated from the culture flask using a xeno-free cell disassociation media after the fourth passage and were lysed by addition of 0.75 mL of Ribozol for total RNA extraction using Epoch Life Science RNA Spin Columns (Sugar Land, TX, USA). cDNA was subsequently synthesized using Bioline SensiFast cDNA synthesis kit (Meridian Life Science, Memphis, TN, USA). Real-time qPCR was then performed on the microglia cDNA to check for the presence of various genes associated with different microglial activation profiles utilizing the BioRad C1000 Thermal Cycler CFX96 Real-Time System (Hercules, CA, USA). All gene primer sets were run as two technical replicates with the use of the SYBR Green fluorophore (ThermoFisher). Cq values were averaged and standardized to GAPDH expression of the sample and then plotted to allow the comparison of each gene’s expression in culture.

To identify P2X7R single nucleotide polymorphisms, genomic DNA was extracted from the human microglia cells. Whole-exosome sequencing was performed by the NovoGene Corp (Chula Vista, CA). A total of 1.0 μg genomic DNA per sample was used as input material for the DNA library preparation, and sequencing libraries were generated using SureSelect Human All Exon kit (Agilent Technologies, CA, USA) following the manufacturer’s recommendations, and index codes were added to each sample. Captured libraries were enriched in a PCR reaction to add index tags to prepare for hybridization. Products were purified using AMPure XP system (Beckman Coulter, Beverly, USA) and quantified using the Agilent high-sensitivity DNA assay on the Agilent Bioanalyzer 2100 system.

We studied both attached and detached microglia. Attached microglia were grown on 13-mm collagen-coated glass coverslips, and free-floating microglia were scraped from 35-mm plastic tissue culture dishes. In both cases, cells were studied in a recording chamber positioned on the stage of a Nikon inverted microscope and continuously perfused with an extracellular solution (ECS) containing the following (in mM): 140 NaCl, 5.4 KCl, 2 CaCl₂, 33 glucose, and 10 HEPES at pH 7.4. Whole-cell currents were recorded at room temperature with low resistance (2–4 MΩ), lightly fire polished, borosilicate glass electrodes (1B150F, World Precision Instruments, Sarasota, FL), and an Axopatch 200B amplifier (Molecular Devices, San Jose, CA) filled with a solution containing the following (in mM): 155 NaCl, 10 HEPES, and 10 EGTA at pH 7.4. The holding potential was − 60 mV except where noted otherwise. Data were filtered at 5 kHz during acquisition and digitized at 10 kHz using ITC-16 data acquisition hardware (Heka Electronics, Holliston, MA). Drugs were applied using triple-barreled theta glass and a Perfusion Fast-Step SF-77 System (Warner Instruments, Hamden, CT). Current-voltage curves were generated either by measuring peak agonist-gated currents (3 s) at a range of steady holding potentials or by measuring the current caused by a 500-ms ramp of voltage from − 90 to 30 mV.

Repetitive applications of ATP and BzATP produce progressive facilitation of P2X7R-mediated currents [27, 28]. To avoid the confounding effect of facilitation on interpretation of concentration-response curves, we applied agonists no more than twice to single cells. We measured the peak current density (pA/pF) for 3-s applications of multiple concentrations of ATP and BzATP, and then we pooled the respective results to yield an average current density for each concentration. These data were plotted as log(agonist concentration) versus current density using Prism 7 (GraphPad, La Jolla, CA) and fit by nonlinear regression to calculate the concentration of agonist giving a half-maximal response (i.e., the EC₅₀).

Human microglia, grown overnight on 13-mm collagen-coated glass coverslips, were incubated for 30 min in normal extracellular solution (+/− Ca²⁺) containing 5 μM Fluo-4-AM and 0.02% (w/v) Pluronic F-127 at room temperature, washed free of the reagents, and left for 30 min at 37 °C to permit de-esterification. Then, single coverslips were transferred to the 14-mm microwell of a MatTek Co (Ashland, MA) glass bottom culture dish positioned on the stage of an Olympus IX70 inverted microscope and visualized (excitation 494 nm, emission 514 nm) using a × 20 objective (0.75 N.A.). Images were captured at a rate of five frames per second using MicroManager [29]. Each image contained 30–50 cells defining regions of interest, and each experiment was repeated at least ten times. Data traces show the fold change in fluorescence over baseline after background subtraction (F/F₀).

Human microglia were grown in 35-mm dishes for 24–72 h. The cells were incubated with or without the P2X7R antagonist A438079 (50 μM) in normal extracellular solution for 1 h at 37 °C. To measure phagocytosis, the microglia were subsequently incubated with 20 μg/mL pHrodo Red E. coli BioParticles Conjugate with or without BzATP (300 μM) in ECS for 16 h at 37 °C. After washing, microglia were scraped and pelleted at 750 rpm for 5 min. The cell pellet was resuspended with 1× FAM-YVAD-FMK (Caspase-1 FLICA; Immunochemistry Technologies, Bloomington, MN, USA) and incubated at 37 °C for 90 min, with gentle mixing every 10 min. The microglia were pelleted, washed two times with Apoptosis Wash Buffer (Immunochemistry Technologies, Bloomington, MN), and resuspended in fresh normal extracellular solution. Cells were then added directly to an inverted epifluorescence microscope, pHrodo Red E. coli BioParticles were detected using 596/615 excitation and emission wavelengths, and Caspase-1 FLICA was detected using 488/510 excitation and emission wavelengths.

Human microglia were plated onto collagen-coated coverslips for 24–72 h. The cells were incubated with or without BzATP (300 μM) for 16 h. To measure intracellular pH, the microglia were incubated with pHrodo Green AM Intracellular pH Indicator in ECS for 30 min at 37 °C. Cells were washed two times in ECS and mounted onto an inverted epifluorescence microscope. pHrodo Green was detected using 488/510 excitation and emission wavelengths.

Microglia cells were seeded in 96-well plates at a concentration of 7 × 10⁴ cells/well and incubated overnight. Cells were primed for 4, 6, or 24 h with 1 μg/ml or 10 μg/ml LPS in ECS and subsequently stimulated for 30 min with BzATP (300 μM) or ATP (5 mM) or 3 h with nigericin (20 μM) in 100 μl ECS at 37 °C. Microglia were also primed with 20 μg/mL pHrodo Red E. coli BioParticles Conjugate ± BzATP (300 μM) for 24 h. THP-1 monocytes were seeded in 96-well plates at a concentration of 5 × 10⁴ cells/well and incubated overnight. THP-1 monocytes were primed with 1 μg/mL LPS for 4 h followed by stimulation for 30 min with BzATP (300 μM) or 3 h with nigericin (20 μM) in 100 μl ECS at 37 °C. Supernatants and lysates (collected using lysis buffer containing the following (in mM): 150 NaCl, 25 HEPES, 5 EDTA, 1% Triton X-100, and 1× SIGMAFAST™ Protease Inhibitor Tablets (Sigma), pH 7.4) were collected and kept frozen at − 20 °C until analysis. Pro and mature IL-1β was evaluated with ELISA using the R&D Systems DuoSet kit (Cat # DY201-05) and the Invitrogen IL-1 beta Human Uncoated ELISA Kit (Cat # 88-7261-88) according to the manufacturer’s protocol. Total IL-18 was evaluated with ELISA using the R&D Systems Human Total IL-18/IL-1F4 Quantikine ELISA Kit (Cat # DL180) according to the manufacturer’s protocol. Developed plates were read on a Biotek Neo Alpha Plate Reader plate reader with Gen5 software (BioTek Instruments, Winooski, VT).

Microglia dye uptake was assayed using the dyes YO-PRO-1 (14 μM), YOYO-1 (10 μM) carboxyfluorescein (0.5 mM), Lucifer yellow (0.5 mM), or ethidium (10 μM). Cells were grown on collagen-coated coverslips and washed in normal ECS. Cells were incubated with 300 μM BzATP in ECS with and without the dyes for 15 min at 37 °C in a humidified atmosphere containing 5% CO₂. In some experiments, cells were pre-incubated for 30 min with the following compounds: A804598 (20 μM), A438079 (50 μM), BX430 (10 μM), Tannic Acid (20 μM), DIDS (100 μM), carbenoxolone (20 μM), probenecid (5 mM), ¹⁰Panx1 peptide (300 μM). In Ca²⁺ free experiments, microglia were pre-incubated with BAPTA-AM (10 μM) for 60 min followed by addition of YO-PRO-1 ± BzATP in Ca²⁺ free ECS containing EDTA (1 mM). In experiments using nigericin, cells were incubated with YO-PRO-1 and nigericin (20 μM) for 30 min at 37 °C. In Cl⁻-free experiments, microglia cells were treated in solution containing the following (in mM): 140 Na gluconate, 5 K gluconate, 5.5 glucose, and 10 HEPES, pH 7.4. In the time-course experiments, YO-PRO-1 and ethidium fluorescence were measured after BzATP stimulation every 30 s for 30 min. The dye uptake was measured by fluorescence microscopy using an inverted epifluorescence microscope (Eclipse TE2000, Nikon) fitted with a CCD camera. YO-PRO-1, YOYO-1, carboxyfluorescein, and Lucifer yellow were measured using 488/510 excitation and emission wavelengths. Ethidium was measured using 596/615 excitation and emission wavelengths.

Microglia were seeded in 96-well plates at a concentration of 7 × 10⁴ cells/well and incubated overnight. Microglia were then treated with BzATP (300 μM), ATP (5 mM), or nigericin (20 μM) for 30 min at 37 °C. Untreated microglia served as the negative control while lysed cells served as the positive control. Fifty microliters of cell supernatant was collected and used to detect LDH activity with the CytoTox96 Non-Radioactive Cytotoxicity Kit (Promega, Madison, WI, USA) according to the manufacturer’s instructions.

In all fluorescence microscopy observations, cells were also observed with clear field illumination and 25–150 cells were present in each microscope field studied. Imaging acquisition and data analysis were performed with the software package ImageJ. We used the following formula to measure the corrected total cell fluorescence (CTCF) in relative fluorescence units (RFU):where “whole cell signal” equals the sum of the intensity of the pixels for one cell, area equals the number of pixels defining the cell, and “background signal” equals the average signal per pixel for a region devoid of cells but close to the cell of interest [30].

Data were analyzed using GraphPad Prism and reported as mean ± s.e.m. Student’s t test (for paired or unpaired samples as appropriate) and analysis of variance with Tukey post-test were used for statistical analysis. p < 0.05 was accepted as a significant difference.

Real-time quantitative PCR was performed to check for the presence of genes associated with different activation states (Fig. 1). First, we sought to confirm the identity of the cultured cells by measuring expression of IBA-1. IBA-1 is a cytoplasmic protein that is exclusively expressed in brain microglia. We saw abundant expression in our cultured cells, confirming their identity. Then, we focused on three genes (interferon gamma (IFNγ), toll-like receptor 4 (TLR4), and tumor necrosis factor alpha (TNFα)) to determine the phenotype of the cultured microglia. Microglia are highly reactive cells that have the ability to express both classical pro-inflammatory and phagocytotic markers. TLR4 is a pattern recognition receptor for LPS, which triggers transcription of both IL-1β and IL-18 precursors. The fact that we saw no measurable TLR4 gene expression in our cultured human microglia suggests an anti-inflammatory phenotype. Next, we measured expression of two additional genes, IFNγ and TNFα, that are associated with the pro-inflammatory phenotype and again saw low expression. Finally, we measured expression of genes typically associated with the phagocytosis phenotype and saw robust expression of Arg-1 and YM1/2. Altogether, these data suggest that primary human microglia adopt an anti-inflammatory and phagocytotic state under the culture conditions used in our experiments.

We used whole-cell voltage-clamp electrophysiology to explore the presence of functional P2X receptors in cultured human microglia. ATP (5 mM) induced a biphasic inward current at a holding potential of − 60 mV whose amplitude progressively grew during a 3-s application (Fig. 2a) This “facilitating” phenotype closely matched that predicted for a uniform population of P2X7Rs, which are the purinergic channels most closely associated with DAMP-mediated responses [32]. The agonist-gated peak current amplitude varied as a function of the holding potential and reversed direction at 0 mV (Fig. 2b, c), as expected for non-selective cation current through an ionotropic P2X receptor [33]. The average reversal potential measured from 13 cells was 0.9 ± 1.5 mV. We saw no evidence of an initial phase of a rapidly desensitizing inward current expected from P2X1Rs and P2X3Rs, and we failed to record membrane current in response to 100 μM ATP thus ruling out P2X2Rs, P2X4Rs, and P2X5Rs.

We measured the dependence of the current amplitude on the concentration of ATP and the higher potency agonist, BzATP, and calculated EC₅₀s of 4.7 mM and 60.8 μM (Fig. 2d), respectively, which are close to those previously reported for recombinant human P2X7Rs determined in the presence of a physiological concentration of Ca²⁺ [34, 35]. We then studied the effects of selective receptor antagonists using 5 mM ATP as agonist. We used this concentration of ATP because it evoked currents with easily measured peak amplitudes and produced full facilitation in a reasonable amount of time. We found that inward current gated by 5 mM ATP was inhibited by preincubating the cells for 2 min in the competitive P2X7R antagonist A438079 (50 μM) [36] and the non-competitive P2X7R antagonist A804598 (20 μM) [37] (Fig. 2e, f). The inhibitions were highly significant but incomplete at the concentrations of ATP and antagonists tested; we did not test higher antagonist concentrations prone to non-selective effects. The remaining ATP-gated current was not inhibited by the selective P2X4R antagonist BX430 [38], confirming the absence of a contribution of P2X4Rs (Fig. 2f). Further, we pre-incubated cells in 10 μM ivermectin, a positive allosteric modulator of P2X4Rs [39], and continued to see no current in response to 100 μM ATP (Additional file 1: Figure S1a), providing additional evidence that cultured human microglia do not express functional P2X4Rs in their cell surface membrane. In contrast, we recorded significant potentiation by ivermectin of the current gated by a saturating concentration of ATP (10 mM; Additional file 1: Figure S1a, b), which supports the hypothesis that ivermectin is also a positive allosteric modulator of human P2X7Rs as previously reported for monocyte-derived human macrophages [40].

Next, we took a closer look at the current facilitation which is a hallmark property of P2X7Rs [27]. We applied ATP once every 15 s (Fig. 3a) and saw a progressive increase in peak inward current amplitude that grew fourfold over the course of 60 3-s applications (Fig. 3b). As expected for a P2X7R-mediated response [28], increasing the length of agonist exposure in a single volley from 3 to 15 s produced a similar degree of facilitation with a fewer number of repetitive ATP applications (Fig. 3c, d).

Taken together, the shape of the current, the reversal potential near 0 mV, the overall insensitivity to ATP and BzATP as indicated by their moderately high EC₅₀ values, the markedly higher potency of BzATP by comparison to ATP, the inhibition by P2X7R selective antagonists, the lack of effect of ivermectin on micromolar concentrations of ATP, and the pronounced current facilitation all suggest that cultured human microglia express a uniform population of functional P2X7Rs.

Ca²⁺ triggers a number of downstream signaling pathways in glia [41]. To determine if the P2X7R is a significant source of Ca²⁺ in cultured adult human microglia, we visualized changes in [Ca²⁺]_i using Fluo-4-AM. Microglia grown on coverslips were incubated with 5 μM Fluo-4-AM for 30 min, followed by 30 min in dye-free medium to allow de-esterification. Unstimulated microglia bathed in normal extracellular medium showed a dim but measurable level of fluorescence (excitation 494 nm: emission 514 nm), indicating a low resting [Ca²⁺]_i. Fluorescence rose quickly to a plateau in response to 300 μM BzATP (black trace, Fig. 4) signifying a persistent rise in [Ca²⁺]_i during constant agonist exposure. The increase in [Ca²⁺]_i was unaffected by addition of the non-selective Ca²⁺ channel blocker, cadmium (200 μM), to the extracellular medium (red trace). In contrast, removing extracellular Ca²⁺ completely abolished the fluorescence change (blue trace). Taken together, these results show that Ca²⁺ current through the P2X7R pore is an absolute requirement for the ATP-gated change in [Ca²⁺]_i in cultured adult human microglia.

LPS is a pro-inflammatory stimulus that causes a 2 to 4-fold increase in the P2X4R-mediated response of immortalized mouse microglial cells without affecting P2X7R currents [42, 43]. We sought to determine if a similar pattern occurs in humans. Overnight treatment with LPS caused a 2.5-fold increase in peak current amplitude (control, 2.4 ± 0.2 pA/pF, n = 54; LPS, 6.0 ± 1.7 pA/pF, n = 14) gated by 10 mM of ATP (Fig. 5a) but had no effect on the time course of facilitation (Fig. 5b). However, LPS did not result in a change in the shape of the ATP-gated current, and we saw no indication of a fast spike of current as expected from activation of a population of P2X4Rs. Further, LPS did not render cultured human microglia sensitive to a concentration of ATP (100 μM) sufficient to maximally activate P2X4Rs, suggesting that LPS does not upregulate this receptor in humans. The ability of LPS to increase P2X7R-mediated current density is interesting in light of the lack of measurable expression of the classical LPS receptor, TLR4 (see Fig. 1), suggesting the presence of non-canonical LPS receptors for this action in cultured human microglia [44].

Because phagocytosis is a vital function of microglia during postnatal development [1], we sought to determine if human microglia retain this function in culture. Towards this end, we incubated cells for 16–24 h in ECS containing pHrodo Red E. coli bioparticles that fluoresce in the acidic environment of a phagosome. We saw strong red fluorescence inside cells, thus demonstrating phagocytosis (inset, Fig. 5d). We measured ATP-currents in fluorescing cells and found that phagocytosis was accompanied by a 3-fold enhancement of peak current density measured from the first application of 10 mM ATP by comparison to cells not exposed to the E. coli bioparticles (Fig. 5c–e); these enhanced currents remained cation non-selective (Additional file 2: Figure S2a). Surprisingly, the phagocytic microglia displayed two distinct current profiles with repetitive activation. Some cells showed a typical pattern of current facilitation while others (11 of 21 for the data of Fig. 5d) showed a gradual decrease in peak current density with repetitive applications (rundown) that eventually plateaued to a steady-state (Fig. 5f). A typical recording of this current rundown is shown in Additional file 2: Figure S2b. At present, the reason why some cells show facilitation while others show rundown is unknown, as we saw no other obvious differences in the behavior of these cells. HEK293 cells transiently expressing rat P2X7Rs occasionally rundown with repetitive applications of ATP as a specific tyrosine in the pore-lining transmembrane domain is dephosphorylated [45], and it is possible that phagocytosis of E. coli in human microglia in some cases stimulates recruitment of a phosphatase to the P2X7R signaling complex.

P2X7Rs act as scavenger receptors that promote phagocytosis by directly binding apoptotic cells and bacteria to their extracellular domain in the absence of receptor stimulation by ATP [46]. The scavenger receptor function of the P2X7R depends on its tight association with non-muscle myosin IIA, which only occurs in the resting conformation [47]. Upon activation of the P2X7R with ATP, non-muscle myosin is dissociated from this complex and phagocytosis is significantly reduced [48]. To determine if activation of P2X7Rs blocks phagocytosis in cultured human microglia, we measured phagocytosis of red-fluorescent E. coli bioparticles in the absence and presence of BzATP (300 μM). We measured significant inhibition of phagocytosis that was prevented by pre-incubation with the P2X7R antagonist A438079 (50 μM; Fig. 6a–c). The ability of BzATP to inhibit pH-sensitive E. coli fluorescence was not caused by a global change in cytoplasmic pH which remained unchanged after application of BzATP (Additional file 3: Figure S3a). Thus, our data support the contention that unstimulated P2X7Rs act as scavenger receptors in human tissues.

In human microglia, we measured strong activation of caspase-1, the enzyme responsible for maturation of IL-1β and IL-18, by 300 μM BzATP which was blocked by antagonizing the P2X7R (Fig. 6a, d, e). Surprisingly, the E. coli-positive phagocytic microglia were entirely negative for activated caspase-1 (i.e., co-localization was never observed between the two signals; see Fig. 6a). These results suggest that activation of the P2X7R causes human microglia to limit anti-inflammatory pathways that support engulfment of pathogens and instead promote inflammatory pathways that activate caspase-1. However, despite activation of caspase-1, we were unable to detect either production or release of IL-1β from cultured human microglia primed with LPS or E. coli particles and stimulated with BzATP, ATP, or the K⁺/H⁺ ionophore nigericin. As a positive control, we detected robust release of IL-1β from the immortalized human monocyte cell line (THP-1 cells) primed with LPS and stimulated with either BzATP or nigericin (Additional file 3: Figure S3b). Since caspase-1 is also responsible for maturation of IL-18, we tested whether activated microglia were capable of releasing this pro-inflammatory cytokine. Interestingly, microglia primed with LPS and activated with either BzATP or nigericin were unable to stimulate the release of IL-18, despite having intracellular IL-18 present in their lysates (Additional file 3: Figure S3c). Previous reports in human monocytes found that the homozygous ⁴⁹⁶Glu to Ala (⁴⁹⁶Ala/Ala) single nucleotide polymorphism (SNP) in the P2X7R impairs ATP-induced IL-1β and IL-18 release [49, 50]. We used an external vendor (Novogene Corp, Chula Vista, CA) to perform whole exosome sequencing to determine if either of our donors carry this mutation; both were heterozygous at ⁴⁹⁶Glu/Ala. Published work shows that human monocytes from heterozygotic donors phagocytose fluorescent beads and undergo ATP-evoked membrane permeabilization, suggesting that ⁴⁹⁶Glu/Ala monocytes retain the wild-type phenotype [51]. We found that ATP triggers inward current blocked by P2X7R antagonists in microglia from heterozygotes (see Fig. 2f), demonstrating functional P2X7Rs in the plasma membrane of our cultured cells. Thus, we conclude that the inability of ATP to trigger release of inflammatory cytokines from cultured human microglia does not result from a loss-of-function P2X7R phenotype. Further, our data show that P2X7R-mediated caspase-1 activation in cultured human microglia does not trigger the release of IL-1β or IL-18.

In cultured murine microglia, activation of P2X7Rs causes blebbing of the membrane, leading to shedding of microvesicles that contain phosphatidylserine on their outer leaflet and IL-1β inside [52]. We used confocal microscopy to image changes in the cell surface membrane and found that activation of the P2X7R with BzATP does not cause bleb formation in human microglia (Additional file 4: Video S1). As a positive control, we identified robust BzATP-induced blebbing in HEK293T cells transfected with human P2X7Rs (Additional file 3: Figure S3d). Moreover, microglia cells did not undergo phosphatidylserine switch in the presence of BzATP, even after stimulation with E. coli particles (Additional file 3: Figure S3e). Thus, the inability of BzATP to stimulate blebbing and phosphatidylserine flip suggest that P2X7Rs do not trigger apoptosis of cultured human microglia cells.

The immediate effect of activating P2X7Rs is the increased flow of Na⁺, K⁺, and Ca²⁺ across the cell surface membrane. The net result is depolarizing current at physiological membrane potentials [33]. A second effect occurs as the membrane becomes permeable to larger molecules (< 900 Da), initiating a number of downstream sequela including inflammation and cell death [53, 54]. To determine if cultured human microglia undergo permeabilization, we measured the change in whole-cell fluorescence that accompanies uptake of the large (630 Da) carbocyanine nucleic acid stain YO-PRO-1. We measured YO-PRO-1 fluorescence before and during 15–30-min applications of 300 μM BzATP and saw a time-dependent increase in total cell fluorescence (Fig. 6a–c). We noted similar results using ethidium, another nucleic acid stain (394 Da; Additional file 5: Figure S4a, b). YO-PRO-1 uptake was strongly inhibited by the P2X7R selective blockers A438079 (50 μM) and A804598 (20 μM) but not by the P2X4R antagonist BX430 (10 μM) (Fig. 7a–c). Activated P2X7Rs were unable to permeate the larger cationic dye YOYO-1 (1270 Da, Additional file 5: Figure S4c), indicating that the dye uptake pathway had a molecular size cutoff point. Moreover, the dye uptake pathway was non-lytic as neither BzATP nor ATP stimulated LDH release from the microglia (Additional file 5: Figure S4d).

ATP triggers uptake of cationic and anionic dyes through independent pathways in Raw 264.7 cells, a macrophage-derived mouse cell line expressing the P2X7R [55, 56]. YO-PRO-1 and ethidium are cations, and as noted above, we measured robust uptake of both dyes in cultured human microglia. In contrast, 300 μM BzATP did not trigger significant uptake of the fluorescent anions Lucifer yellow (457 Da; 500 μM) or carboxyfluorescein (376 Da; 500 μM) (Fig. 7d, e). As a positive control [57], we measured robust ATP-induced Lucifer yellow and carboxyfluorescein uptake in J774 mouse macrophages (Fig. 7f). Taken together, our data show that cultured human microglia express an ATP-dependent pathway for the passage of large organic cations across the cell surface membrane but lack the conduit for large anions present in murine immune cells.

The identity of the cation-selective pathway responsible for permeabilization is controversial. Strong evidence support direct passage of YO-PRO-1 through the P2X7R pore [53, 58], while other reports suggest the involvement of downstream proteins [59‐61]. The most prominent candidate is pannexin-1, a hemi-channel permeable to YO-PRO-1 [62] and implicated in permeabilization of murine microglial BV-2 cells [63] and release of IL-1β from murine macrophages [64]. We measured whole-cell fluorescence in the presence of a range of pannexin-1 inhibitors (20 μM carbenoxolone, 5 mM probenecid, and 300 μM ¹⁰Panx inhibitory peptide) and saw no effect of these antagonists on the robust uptake of YO-PRO-1 by human microglia incubated in 300 μM BzATP for 15 min (Fig. 8a). Moreover, carbenoxolone had no effect on the peak amplitude of the fully facilitated ATP-gated current measured using electrophysiology (Fig. 8b, c). Collectively, our results show that both ATP-induced membrane permeabilization and cation channel activity are independent of pannexin-1 channels in cultured human microglia.

A recent report suggests that anoctamin 6 (ANO6; also known as TMEM16F), a Ca²⁺-dependent phospholipid scramblase and Ca²⁺-activated Cl⁻ channel, is required for ATP-induced dye uptake downstream of the P2X7R in primary mouse macrophages and human THP-1 cells [65]. Therefore, we sought to determine the role of ANO6 on BzATP-induced YO-PRO-1 uptake in primary human microglia. We discovered that the nonselective Cl⁻ channel blockers tannic acid (20 μM) and DIDS (100 μM) strongly inhibited YO-PRO-1 uptake (Fig. 9a) without affecting the ATP-gated membrane current (Fig. 9b, c) or the ability of this current to facilitate during repetitive applications (Fig. 9d). The fact that tannic acid and DIDS block YO-PRO-1 uptake but not ATP-gated membrane current supports the hypothesis that the P2X7R channel does not form the physical pore through which large cationic dyes enter the cell.

Next, we investigated the role of Ca²⁺ in mediating membrane permeabilization because ATP increases [Ca²⁺]_i in human microglia (see Fig. 4) and ANO6 is a Ca²⁺-activated Cl⁻ channel. We incubated microglia in 10 μM BAPTA-AM for 30 min, and then tested the ability of 300 μM BzATP to permeabilize cells bathed in a Ca²⁺-free solution containing 1 mM EDTA. In control experiments using cultured human microglia transiently expressing the genetic Ca²⁺ indicator, GCaMP5, we confirmed that microglia loaded with BAPTA and bathed in EDTA show no change in [Ca²⁺]_i in response to BzATP (Additional file 6: Figure S5). Despite this fact, we measured strong YO-PRO-1 fluorescence under these conditions (Fig. 9a). The fact that permeabilization persists in the absence of a change in [Ca²⁺]_i argues against a role for ANO6 in the ATP-dependent pathway in cultured human microglia.

In mouse macrophages, P2X7Rs are thought to trigger dye uptake by depleting intracellular K⁺ [66]. We tested the dependence of dye uptake on K⁺ depletion in cultured human microglia using the K⁺/H⁺ ionophore nigericin. Stimulation with 20 μM nigericin for 30 min resulted in robust uptake of YO-PRO-1 (Fig. 9e). The dye uptake was not due to cell lysis since nigericin did not stimulate LDH release (Additional file 5: Figure S4d). Interestingly, nigericin-stimulated dye uptake was inhibited by tannic acid (Fig. 9f, g), demonstrating that the effect of nigericin occurs upstream of the Cl⁻ channel activation.