Stimulation of P2X₇ Receptors Elevates Ca²⁺ and Kills Retinal Ganglion Cells

Abstract

purpose. Retinal ganglion cells are known to express ionotropic P2X₇ receptors for ATP. Stimulation of these receptors in other cells can elevate Ca²⁺ and sometimes lead to cell death. This study asked whether P2X₇ receptor stimulation alters the Ca²⁺ levels and viability of retinal ganglion cells.

methods. P2X₇ agonists were applied to retinal ganglion cells from neonatal rats loaded with fura-2 to examine their effect on intracellular Ca²⁺ levels. The effect of P2X₇ receptor stimulation on cell viability was examined in rat retinal ganglion cells back-labeled with aminostilbamidine.

results. The P2X₇ agonist benzoylbenzoyl adenosine triphosphate (BzATP) led to a large, sustained increase in Ca²⁺. BzATP was >100-fold more effective than ATP at raising intracellular Ca²⁺, when both agonists were applied at 10 μM. The response to BzATP was enhanced threefold by removal of extracellular Mg²⁺, was dependent on extracellular Ca²⁺, and was prevented by brilliant blue G (BBG). BzATP led to a concentration-dependent reduction in the number of cells with a median lethal dose (LD₅₀) of 35 μM. Cell death was prevented by the P2X₇ antagonists BBG and oxidized ATP, but not by 30 μM suramin, consistent with the actions of the P2X₇ receptor. BzATP activated caspases in ganglion cells, but did not lead to membrane blebbing or increased permeability to Yo-Pro-1. The L-type Ca²⁺ channel blocker nifedipine attenuated cell death, suggesting excessive Ca²⁺ influx contributes to the lethal effects of BzATP.

conclusion. Short-term stimulation of the P2X₇ receptor can raise Ca²⁺ in rat retinal ganglion cells, whereas sustained stimulation of the receptor can kill them.

ATP functions as an extracellular signaling molecule on both neural and non-neural cells. Responses are initiated after binding of ATP to either metabotropic P2Y or ionotropic P2X receptors, and members of both receptor families have been localized to the posterior eye. ^{1 2 3 4} Recent evidence suggests that the P2X₇ receptor contributes to retinal function. Expression of the P2X₇ receptor in Müller cells is upregulated in proliferative vitreoretinopathy, ^{5 6 7} whereas receptor stimulation can modulate the contractility of retinal pericytes. ^{8 9} The receptor is found in both inner and outer retinal neurons, with expression on retinal ganglion cells of the primate ¹⁰ and rat retina. ^{11 12 13} However, the functional consequences of receptor stimulation are less clear. Stimulation of cultured rat ganglion cells with ATP was shown to activate inward currents, ¹⁴ but the lack of specific agents available at the time precluded assignment of the responses to a particular P2X receptor. 

The precise effect of stimulating the P2X₇ receptor on retinal ganglion cell function is hard to predict, as the receptor has been associated with unusual characteristics in other cells. Like other members of the P2X receptor family, stimulation of the P2X₇ receptor can open an ionotropic channel permeable to mono- and divalent cations leading to an elevation of intracellular Ca²⁺. ¹⁵ However, the P2X₇ receptor contains a unique cytoplasmic tail that has been associated with additional channel dilation. ^{16 17} In some peripheral cells, P2X₇ receptor stimulation is also linked to the opening of a large pore of more than 3 nm and to an increased permeability to molecules up to 900 Da, such as the dye Yo-Pro-1. ^{18 19} Pore activation can lead to cell lysis and death and may be linked to the P2X₇ receptor through a large receptor-signaling complex of cytoskeletal and integrin molecules. ²⁰ Although this complex is found in peripheral immune cells, it may not be associated with the P2X₇ receptor in the central nervous system (CNS). ²¹ It is thus unclear whether stimulation of neuronal P2X₇ receptors would have the lethal effects observed in peripheral systems. 

In the present study, we investigated the ability of P2X₇ receptor agonists to elevate intracellular Ca²⁺ levels and kill ganglion cells. Receptor identity was confirmed using several pharmacologic and physiologic approaches, and the mechanism underlying cell death was probed. The findings suggest that the P2X₇ receptor may contribute to ganglion cell function in both physiologic and pathophysiologic ways. 

Portions of this manuscript have been reported in abstract form (Zhang X, et al. IOVS 2002;43:ARVO E-Abstract 2777; Zhang X, et al. IOVS 2003;44:ARVO E-Abstract 5201; Mitchell CH, et al. IOVS 2004;45:ARVO E-Abstract 2083). 

Materials and Methods

Retinal Cell Culture and Labeling of RGCs

As nearly all of the rat’s retinal ganglion cells project to the contralateral superior colliculus, retrograde transport of marker from the superior colliculus is an efficient way to label and analyze ganglion cells from within the retina. ^{22 23} Pups aged postnatal day (PD) 2 to PD 6 from untimed pregnant Long-Evan rats (Jackson Laboratory Inc., Bar Harbor, ME) were back-labeled by the injection of gold tracer derivative aminostilbamidine (Molecular Probes, Eugene, OR) based on standard protocols. ²⁴ Pups were anesthetized with an intraperitoneal (IP) injection of 50 mg/kg ketamine and 5 mg/kg xylazine, an incision was made to expose the skull, and a 1-mm hole was drilled through the skull to expose the cortex overlying each superior colliculus. With a syringe (Hamilton, Reno, NV) affixed to a micromanipulator, a needle was inserted 0.8 mm lateral from the midline and 0.8 mm anterior to Bregma’s line, and 2.5 μL dye was delivered to each side at a depth of 2 and 1 mm. The needle was retracted after a delay of 2 minutes, to allow dye absorption, and the wound was closed with two to three sutures. Preliminary examination of labeled retinal wholemounts confirmed an even distribution of dye, showing dense staining of cells 2 days after injection and no further increase in the number of labeled cells in subsequent days. Consequently, retinas containing labeled ganglion cells were dissociated 2 to 6 days after injection. Animals were killed by IP injection of 50 mg/kg ketamine and 5 mg/kg xylazine followed by an overdose, in accordance with University of Pennsylvania Institutional Animal Care and Use Committee (IACUC)–approved protocols and the ARVO Statement on the Use of Animals in Ophthalmic and Vision Research. 

Retinal Culture

The retina was dissected from each globe, washed in sterile Hanks’ balanced salt solution (HBSS; Invitrogen Corp., Carlsbad, CA), and incubated in HBSS containing activated papain (4.5 U/mL; Worthington Biochemical Corp., Lakewood, NJ) for 12 minutes at 37°C. Retinas were washed twice and triturated 50 times with a 1-mL glass pipette, to dissociate the cells. Cells were plated onto twelve 12-mm coverslips previously coated with poly-l-lysine. The basic growth medium contained enzyme-inhibiting medium (Neurobasal; Invitrogen Corp.) with 2 mM glutamine, 100 μg/mL gentamicin, 0.025 mL/mL B27 supplement (all Invitrogen Corp.), 0.7% methylcellulose (Stemcell Technologies Inc., Vancouver, BC, Canada), and 2.5% rat serum (Cocalico Biologicals Inc., Reamstown, PA). Retinal cells were incubated at 37°C with 5% CO₂. 

Cell Viability Studies

Drugs were added to the culture medium at the time when cells were plated onto coverslips. After incubation for the indicated time, coverslips were mounted on a microscope (Eclipse E600; Nikon, Tokyo, Japan) equipped for epifluorescence and the fluorescent cells (360 ± 40 nm excitation, >515 nm emission) present in 80 central fields were counted with a 40× objective. All counts were performed in a masked fashion. In experiments involving pretreatment, inhibitors were added to the medium at the time of plating. After preincubation at 37°C, stock concentrations of benzoylbenzoyl ATP (BzATP) were added directly to the cells to give the final concentration shown. 

Intracellular Ca²⁺ Measurements

Unlabeled retinal ganglion cells grown on coverslips for 24 hours were loaded with 10 μM fura-2 AM and 2% pluronic F-127 (Molecular Probes) for 60 to 90 minutes at room temperature, rinsed, and maintained in fura-2-free solution for 30 minutes before data acquisition began. The coverslips were mounted on an inverted microscope (Diaphot; Nikon) and visualized with a 40× objective. Preliminary experiments using cells labeled with aminostilbamidine dye demonstrated that all bright, granulated cells with axonal processes were fluorescent, allowing individual unlabeled cells to be identified on morphologic criteria. Additional trials using ganglion cells purified with the panning procedure confirmed that >99% of cells possessing these characteristics were labeled, although this morphologic criteria did not apply to all labeled ganglion cells. To obtain Ca²⁺ measurements, the field was alternatively excited at 340 and 380 nm with a scanning monochromator, and the fluorescence emitted >520 nm from a region of interest surrounding individual retinal ganglion cells was imaged with a charge-coupled device (CCD) camera and analyzed (all Photon Technologies International, Inc., Lawrenceville, NJ). Cells were perfused with a control solution at the start of Ca²⁺-imaging experiments containing (in mM) 105 NaCl, 4.5 KCl, 2.8 Na HEPES, 7.2 HEPES acid, 1.3 CaCl₂, 0.5 MgCl_2, 5 glucose, and 75 mannitol (pH 7.4). Drugs were dissolved into the control solution. Calibration was performed separately on each cell after the experiment, by using standard techniques. ²⁵ Cells were perfused with 5 μM ionomycin in high-Ca²⁺ solution followed by ionomycin in Ca²⁺-free solution, including 5 mM EGTA (pH 8.0). The 340/380-ratio was converted to Ca²⁺ concentration using the method of Grynkiewicz et al., ²⁵ as previously described. ²⁶ All Ca²⁺ measurements were performed at room temperature. 

Yo-Pro-1 Uptake

Mixed retinal cultures containing ganglion cells labeled with aminostilbamidine were cultured for 18 to 24 hours, as described earlier. Coverslips were placed into a four-well culture plate containing 1 mL phosphate-buffered saline (PBS) with 50 μM BzATP and 0.1 μM Yo-Pro-1 dye (Molecular Probes). Coverslips were incubated in the dark at room temperature for 10 minutes, placed on a glass slide, mounted, and observed with a fluorescence microscope (Eclipse E600; Nikon) with a ×40 objective. The Yo-Pro-1 dye was excited at 480 ± 20 nm with emission >515 nm, and aminostilbamidine was excited at 360 ± 20 nm with >515 nm emission. There was no overlap between the dyes with the filter sets used. Images of each field were taken with both filter sets with a 3-CCD digital camera (Toshiba America, Irvine, CA) and analyzed off-line (Image Pro Plus software; Media Cybernetics, Silver Spring, MD) Fluorescence from Yo-Pro-1-stained cells was assigned a red pseudocolor and fluorescence from the aminostilbamidine channel a green pseudocolor, to facilitate analysis. 

Caspase Activation Assay

Caspase activation was detected in purified ganglion cells that were obtained using the panning procedure described later. BzATP was added to the medium at the time of plating, and cells were incubated at 37°C with 5% CO₂ for 24 hours. After the CaspACE FITC-VAD-FMK In Situ Marker (Promega Corp., Madison, WI; final concentration 10 μM) was added, the cells were incubated at 37°C with 5% CO₂ for 20 minutes. Coverslips were washed twice with PBS, placed on a slide, mounted, and coverslipped with antifade mounting medium. Care was taken to keep the coverslips protected from light in all the steps. Purified ganglion cells were observed with a fluorescence microscope (Eclipse E600; Nikon). Apoptotic cells were stained with a green fluorescein marker (480 ± 20 nm excitation, >515 nm emission; CaspACE FITC-VAD-FMK In Situ Marker), whereas ganglion cells labeled with aminostilbamidine fluoresced yellow-green (360 ± 20 nm excitation, >515 nm emission). Dye overlap was minimal. Images of each field were taken using both filter sets and analyzed as above. 

Ganglion Cell Panning

Ganglion cells were purified by using the panning procedure, according to published methods. ^{27 28 29} Neonatal rat retinas (PD 7-12) were dissected and incubated at 37°C for 30 minutes in HBSS containing 15 U/mL papain, 0.2 mg/mL dl-cysteine and 0.004% DNase I (Worthington/Cooper, Lakewood, NJ). The tissue was triturated in HBSS with 1.5 mg/mL ovomucoid (Worthington/Cooper), 1.5 mg/mL BSA and 0.004% DNase I, centrifuged at 200 g for 11 minutes at room temperature, and rewashed with 10 mg/mL ovomucoid-BSA solution. After centrifugation, cells were resuspended with PBS containing 0.2 mg/mL BSA and 5 μg/mL insulin and filtered through a mesh (Nitex; Small Parts Inc., Miami Lakes, FL). Cells were incubated with rabbit antimacrophage antibody (1:75; Accurate Chemical, Westbury, NY), incubated in a 100-mm dish coated with goat anti-rabbit IgG antibody (1:400; Jackson ImmunoResearch Laboratories Inc., West Grove, PA). Nonadherent cells were removed to a second Petri dish coated with goat anti-mouse IgM (1:300; Jackson ImmunoResearch Laboratories Inc.) and anti-Thy 1.1 antibody (from hybridoma T11D7e2; American Type Culture Collection, Manassas, VA). After 30 minutes, nonadherent cells were washed off and incubated with 0.125% trypsin for 8 minutes at 37°C. Fetal bovine serum (30%) in neural basal medium was used to stop the digestion, and the cells were centrifuged and plated as just described, on coverslips coated with poly-l-lysine and laminin. 

Data Analysis

Data are presented as mean ± SEM. Significance was evaluated using a one-way ANOVA followed by a Student-Newman-Keuls test when more than two variables were present or an unpaired Student’s t-test when only two variables were present. For cell viability studies, the number of experiments, n, represents the number of coverslips from which 80 fields were measured and averaged. All results were normalized to the mean control level for that day’s matched set of experiments, to control for variation in plating efficiency. In Ca²⁺ experiments, n refers to the number of responses tested. The BzATP concentration–response curve was fit with a standard exponential function y = y ₀+ae ^{(−bx )} on computer (SigmaPlot software; SPSS Science, Inc., Chicago, IL). The percentage block of Ca²⁺ elevations is defined as 100 · (a − b)/a, where a is the response in control conditions, and b is the response in experimental conditions. All materials are from Sigma-Aldrich (St. Louis, MO) unless otherwise indicated. 

Results

P2X₇ Receptor Stimulation and Ca²⁺

Previous work has demonstrated that activation of the cloned P2X₇ receptor opens up a channel permeable to mono- and divalent cations. ¹⁸ Initial functional assays for the P2X₇ receptor on retinal ganglion cells thus examined the effect of stimulation on levels of intracellular Ca²⁺. The P2X₇ agonist BzATP led to a rapid increase in intracellular Ca²⁺, and the level remained elevated for the 2-minute duration of agonist application, although a small relaxation was frequently observed (Fig. 1A) . Over 80% of the cells examined responded robustly to BzATP. The increase in Ca²⁺ was evenly distributed across the cells within the spatial and temporal resolution of the recording. When the magnitude of the response was quantified, application of 50 μM BzATP for 2 minutes in control perfusion solution increased peak Ca²⁺ levels by 557 ± 85 nM (n = 9). Although levels of Ca²⁺ declined quickly after removal of BzATP, treatment led to a small long-term elevation in Ca²⁺, with concentrations 5 minutes after drug washout remaining 20 ± 8 nM higher than levels before application of BzATP (n = 9). A second 2-minute application of BzATP after a 5-minute wash triggered another increase in Ca²⁺, slightly smaller than the first, with a mean peak increase of 477 ± 97 nM (n = 9). 

The cloned rat P2X₇ receptor is blocked by extracellular Mg²⁺, with an IC₅₀ of 0.47 mM. ³⁰ Because the control solution used contained 0.8 mM Mg²⁺, experiments were repeated in the absence of Mg²⁺ to determine whether the cation inhibited the response. The effect of BzATP was increased nearly threefold in the absence of extracellular Mg²⁺, with levels rising by 1438 ± 145 nM in cells exposed to 50 μM BzATP (n = 9, P < 0.0005 versus BzATP with Mg²⁺; Fig. 1B ). Although removal of BzATP led to a prompt reduction, Ca²⁺ levels remained considerably elevated compared with those in Mg²⁺-containing solutions, with concentrations 156 ± 22 nM higher than baseline 5 minutes after drug washout (n = 9, P < 0.0005 versus BzATP with Mg²⁺). A second addition of BzATP raised levels by 606 ± 101 nM. 

In contrast to other cloned P2X receptors, the P2X₇ receptor is more responsive to BzATP than to ATP. ¹⁵ In retinal ganglion cells, BzATP was considerably more effective at elevating the intracellular Ca²⁺ than ATP. Figure 1Cillustrates that 10 μM BzATP elevated Ca²⁺ levels more than did 100 μM ATP. Although the magnitude of the responses and their ratio varied, BzATP produced a larger elevation than ATP in all 12 cases in which equal concentrations of the agonist were used. BzATP increased Ca²⁺ 112-fold compared with ATP, when both agonists were used at 10 μM (n = 7); 62-fold at 30 μM (n = 3); and 23-fold at 100 μM (n = 2). The Ca²⁺ levels were determined during the final 10 seconds of agonist application to ensure that this reflected a plateau response consistent with the stimulation of the P2X₇ receptor and not a transient elevation. 

The Ca²⁺ elevations induced by stimulation of the P2X₇ channel are primarily due to the influx of extracellular Ca²⁺, not release from intracellular stores that accompany stimulation of the G-protein-linked P2Y receptors. To determine whether extracellular Ca²⁺ (Ca²⁺ _o) was necessary for the Ca²⁺ increase after BzATP application in retinal ganglion cells, experiments were performed in the absence of Ca²⁺ in the bath. Figure 2Aindicates that in the presence of control solution, 15-second applications of 50 μM BzATP triggered repeatable elevations in intracellular Ca²⁺ that were remarkably consistent (although the peak responses after the brief BzATP application were not as large as those after the 2-minute applications illustrated in Fig. 1 ). This reproducibility allowed the effect of the removal of Ca²⁺ _o to be examined within a single cell. After triggering a response with a brief application of 50 μM BzATP, removal of Ca²⁺ _o led to a small reduction in baseline levels of cell Ca²⁺. Application of BzATP in the absence of Ca²⁺ _o did not elicit a response (Fig. 2B) . However, the agonist triggered a response on return to control Ca²⁺ _o levels, indicating it was the absence of Ca²⁺ _o that prevented an increase in Ca²⁺. The mean response to BzATP was reduced by 97.8% ± 0.44% (n = 6) in the absence of Ca²⁺ _o, when compared with the mean increase of the applications preceding low-Ca²⁺ treatment, with the response falling from 275.8 ± 42.8 nM to only 5.4 ± 0.9 nM (P < 0.0001). 

To further confirm the involvement of the P2X₇ receptor in the BzATP response, the ability of the P2X₇ antagonist brilliant blue G (BBG) to prevent an increase in Ca²⁺ was explored. BBG is frequently used as a specific P2X₇ antagonist, but recent work has indicated it can also prevent the activation of the P2X₅ receptor. ^{31 32} However, the P2X₇ receptor shows a higher affinity for BBG, ³² so the ability of BBG to block the effect of BzATP was examined at several concentrations. Initial exposure to 50 μM BzATP led to a large Ca²⁺ elevation, but BBG blocked the subsequent responses (Fig. 2C) . The inhibitor was almost equally effective over the range examined; 100 nM BBG blocked 94.4% ± 1.6% of the response (n = 4), 1 μM blocked 98.1% ± 2.5% of the response (n = 4), and 10 μM blocked 97.9% ± 2.0% (n = 2). BBG itself led to a small increase in baseline Ca²⁺ levels, and the block by BBG was not reversible. 

P2X₇ Stimulation and Cell Death

Elevations in intracellular Ca²⁺ can contribute to apoptotic cell death in retinal ganglion cells. ^{33 34} As the results reported thus far indicate that BzATP is capable of producing sustained elevations in Ca²⁺, and stimulation of the P2X₇ receptor is known to kill non-neuronal cells, ¹⁶ the effect of the agonist BzATP on cell survival was examined. BzATP was added to freshly dissociated retinal cultures containing labeled ganglion cells. The number of labeled cells remaining after a given interval was determined and compared to that under control conditions, to ascertain specifically the effect of BzATP. There were always fewer ganglion cells present in coverslips incubated with 50 μM BzATP than on control coverslips. Typical trials showed 70 to 80 cells per coverslip in control conditions, with 45 to 55 in the presence of BzATP. When cell survival was determined at different times after agonist addition, it became evident that the proportion of dead cells increased with exposure time to BzATP (Fig. 3A) . 

The ability of BzATP to kill retinal ganglion cells was explored further by examining the effect of agonist concentration on cell survival. Exposing cells to 50, 100, and 500 μM BzATP for 24 hours led to a progressively smaller number of cells remaining on coverslips relative to control conditions (Fig. 3B) . The data were well fit by a single exponential decay, with an asymptotic value of 55%, indicating that less than half the ganglion cells were killed by the agonist at the maximum dose within the 24-hour time frame of the experiments. The fit gave a predicted EC₅₀ of 35 μM BzATP for cell death at the 24-hour point. 

To confirm the contribution of the P2X₇ receptor to cell death, we determined the ability of P2 antagonists to prevent this cell death. Cells were resuspended with the inhibitors, plated as usual, and 50 μM BzATP was added to the cells after 30 minutes. The P2X antagonist oxidized ATP (oATP) prevented cell death at 100 μM (Fig. 3C) . To rule out contributions from other P2 receptors inhibited by oATP, the effect of the antagonist suramin was examined. At 30 μM, suramin had no effect on cell viability when incubated with BzATP (Fig. 3D) ; levels were 69% ± 6% of control with 50 μM BzATP alone and 60% ± 3% of control in the presence of suramin and BzATP. The inhibitor brilliant blue G (BBG) significantly increased cell survival at 100 nM, 1 μM, and 10 μM (Fig. 3E) . Although 24 hours in the presence of 50 μM BzATP reduced cell numbers to only 67% ± 3% of control, survival was increased to 80% ± 5% by inclusion of 100 nM BBG, to 86% ± 4% by 1 μM BBG and to 89% ± 5% survival with 10 μM BBG. Together, the effects of these inhibitors strongly suggest that the stimulation of the P2X₇ receptor is responsible for killing retinal ganglion cells. 

Mechanisms of Cell Death

The mechanisms linking stimulation of the P2X₇ receptor with the death of ganglion cells were examined. Caspase activation was detected in ganglion cells exposed to BzATP using the CaspACE FITC-VAD-FMK in situ marker (Fig. 4A) . Because it was unclear whether a change in aminostilbamidine labeling would accompany apoptosis, experiments were performed with ganglion cells purified with the panning protocol. This purification step had the additional benefit of excluding microglial cells that had previously phagocytosed labeled ganglion cells. After 24 hours in BzATP, 20% of cells were labeled with the CaspACE stain (Fig. 4A) , whereas fewer than 5% of the ganglion cells showed caspase activation in the absence of BzATP. This suggests that BzATP triggers an apoptotic response in retinal ganglion cells consistent with the cell death that accompanied exposure to the agonist. 

Previous studies have indicated that stimulation of the P2X₇ receptor can lead to an increase in the permeability of the plasma membrane to the fluorescent dye Yo-Pro-1 within minutes, ^{19 30} and so the effect of BzATP on Yo-Pro-1 permeability in mixed retinal cultures was examined. Retinal ganglion cells did not show an increase in permeability to Yo-Pro-1 (Fig. 4B) . When inspected 10 to 30 minutes after application of 50 μM BzATP, Yo-Pro-1 uptake was clearly detected in some smaller retinal cells, providing a positive control and consistent with recent reports that retinal microglial cells are permeable to Yo-Pro-1 after exposure to BzATP. ³⁵ However, <5% of ganglion cells showed any uptake of Yo-Pro-1, suggesting that alternative mechanisms were involved triggering apoptosis and death in the neurons. 

As the experiments showed that BzATP triggered large increases in intracellular Ca²⁺ levels and as excess Ca²⁺ is well known to be toxic to neurons, ³⁶ we asked whether increased intracellular Ca²⁺ contributes to cell death. Although removal of extracellular Ca²⁺ for 24 hours was not possible, downstream activation of voltage-dependent Ca²⁺ channels can follow the stimulation of the P2X₇ receptor, ³⁷ and retinal ganglion cells are known to express several voltage-dependent calcium channels including L-type calcium channels. ³⁸ To determine whether influx of Ca²⁺ through L-type calcium channels contributed to cell death, cells were incubated in the presence of BzATP, with and without the L-type calcium channel blocker nifedipine (50 μM). Nifedipine raised ganglion cell survival from 53% ± 5% to 72% ± 6% of control (n = 11 for all), an increase of 40% (Fig. 4C) . This indicates that influx of Ca²⁺ through L-type calcium channels contributes to the death of ganglion cells after stimulation of the P2X₇ receptor. 

Discussion

These results provide the first functional evidence of the presence of the P2X₇ receptor on retinal ganglion cells. The data demonstrate that receptor stimulation can lead to a large and prolonged influx of extracellular Ca²⁺. Sustained activation also has pathophysiological implications and can lead to cell death. This death is not caused by formation of a large pore as in peripheral cells, but depends in part on influx of Ca²⁺ through L-type Ca²⁺ channels. 

Identification of P2X₇ Action

Several observations combine to identify a specific role for the P2X₇ receptor in both the Ca²⁺ elevations and cell death. First, the Ca²⁺ response to BzATP was eliminated by the removal of extracellular Ca²⁺. This implies that the response is not due to the release of Ca²⁺ from intracellular stores that follows stimulation of the G-protein-coupled P2Y receptors. Second, the enhancement of the response after the removal of Mg²⁺ is a characteristic of the P2X₇ receptor. ^{15 39} The magnitude of this enhancement is consistent with the response in cloned rat P2X₇ receptors, where Mg²⁺ removal also led to an approximately threefold increase. ¹⁸ (It is important to stress that all cell viability experiments were performed with 1.8 mM CaCl₂ and 0.8 mM MgCl₂ in the medium, suggesting that cell death can occur in the presence of physiologic levels of divalent cations.) Third, the relative efficacy of BzATP and ATP strongly implicate the P2X₇ receptor. ¹⁵ Whereas the EC₅₀ for BzATP can be greater than (P2X₂, P2X₅), equal to (P2X₃, P2X₄), or less than (P2X₁, P2X₇) that for ATP, the only receptor that shows a larger response to BzATP than to ATP is the P2X₇ receptor. ⁴⁰ Indeed, the >100-fold increase in the response to BzATP versus ATP at 10 μM is similar to that found previously. ⁴⁰ Fourth, the blocks produced by BBG and oATP are consistent with P2X₇ activation. There is currently no specific antagonist for the rat P2X₇ receptor; oATP can produce clear effects independent of the P2X₇ receptor, whereas BBG inhibits human P2X₅ receptors with an IC₅₀ of 530 nM. ^{32 41} However, activity of the P2X₅ receptor is not enhanced by removal of extracellular Mg²⁺. ³² In addition, 100 nM BBG produced only a 5% block of P2X₅ receptors, whereas it had a considerable effect on both the Ca²⁺ response and cell death triggered by BzATP in ganglion cells at this concentration in the present study. Fifth, the inability of suramin to affect cell viability rules out a contribution from many other P2 receptors. At 30 μM, suramin inhibits the P2X₁, P2X₂, P2X₃, P2X₅, P2Y₁, and P2Y₂ receptors. ^{15 42} As the IC₅₀ for the compound at P2X₇ receptors is ∼500 μM, the lack of effect suggests that the contribution from these other P2 receptors is minimal. Sixth, the ability of BzATP to elevate Ca²⁺ in ganglion cells is consistent with its action on P2X₇ receptors and not on P1 receptors, as found in hippocampal cells. ⁴³ Together, these observations imply that stimulation of the P2X₇ receptor is sufficient to raise Ca²⁺ and kill retinal ganglion cells. 

P2X₇ Receptors and Neuronal Cell Death

The contribution of the P2X₇ receptor to neural signaling is currently a matter of some discourse. The receptor has been localized to synaptic regions throughout the brain, ⁴⁴ and physiologic data have suggested that stimulation can modulate neural transmission and alter neurotransmitter release in some regions, including the hippocampus. ^{45 46 47} More recent reports also implicate the receptor in the secondary loss of neurons after spinal cord injury. ⁴⁸ Although the specificity of agonists, antagonists, and antibodies used to identify P2X₇ receptor involvement in these reports has been questioned, ^{41 43 49} the identification of P2X₇ receptors in retinal ganglion cells using both immunologic and molecular techniques approaches ^{10 11 12 13} combined with the improved pharmacology used in the present study, strongly support a role for the P2X₇ receptor in the signaling and pathophysiology of neurons. 

The mechanism connecting P2X₇ receptor stimulation and retinal ganglion cell death merits comment. In some cell types, stimulation of the P2X₇ receptor can lead to an increased permeability of the large dye Yo-Pro-1, membrane blebbing, and cell lysis. ¹⁹ These effects may be due to an association of the P2X₇ receptor with multimeric complexes containing bleb-forming epithelial membrane proteins. ²⁰ However, only peripheral P2X₇ receptors, not those in the brain, are associated with multimeric complexes. ²¹ In an elegant study, Innocenti et al. ³⁵ recently showed that microglial cells were the only retinal cells to increase permeability to Yo-Pro-1 after stimulation with BzATP. This finding is supported by our inability to detect Yo-Pro-1 uptake in ganglion cells in the present study, while detecting the dye in smaller, unlabeled cells. As the retinal microglial cells are immune cells, the increase in their membrane permeability agrees with the general preponderance of pore formation in immune cells. The lack of any pore formation in retinal neural cells in our hands or those of Innocenti et al. ³⁵ is consistent with the lack of multimeric complexes in the CNS. ²¹  

An alternative mechanism leading to ganglion cell death may instead be related to the large influx of Ca²⁺ accompanying stimulation of the P2X₇ receptor and suggests that it is the particular collection of downstream effectors that determines whether stimulation of neuronal P2X₇ receptors are lethal. Although the P2X₇ receptor is somewhat permeable to Ca²⁺, ⁵⁰ it can also lead to a secondary activation of voltage-gated Ca²⁺ channels. After stimulation of the P2X₇ receptor in NG 108-15 cells, influx through the P2X₇ pore itself increased Ca²⁺ by 211 nM, whereas an additional increase of 227 nM Ca²⁺ occurred through N-type Ca²⁺ channels and 189 nM through L-type channels. ³⁷ Examination of currents in rat retinal ganglion cells showed that >60% of the increase in Ca²⁺ after stimulation with ATP was secondary to receptor activation, as removal of extracellular Na⁺ attenuated the response considerably. ¹⁴ Our ability to block cell death with nifedipine agrees with this study and suggests that a secondary influx through Ca²⁺ channels contributed to ganglion cell death after stimulation of the P2X₇ receptor. Nifedipine also reduces cell death associated with N-methyl-d-aspartate (NMDA) receptor stimulation, ³³ implying the over-activation of L-type Ca²⁺ channels may be a common step in the neurotoxic death of retinal ganglion cells. Although the contribution of other Ca²⁺ channels known to be present in these cells is currently being investigated, ³⁸ the present results indicate that recruitment of Ca²⁺ channels contributes to cell death. The use of mixed retinal cultures for both Ca²⁺ measurements and cell viability experiments means that the involvement of other cell types cannot be excluded. However, the ability of BzATP to activate caspase in purified ganglion cells (Fig. 4A)suggests all necessary components are present within these cells. The difference between the proportion of ganglion cells dying in response to BzATP and those in which Ca²⁺ levels were increased suggests the large, granulated cells chosen for Ca²⁺ measurements reflect a subpopulation of cells expressing the P2X₇ receptor. 

Physiologic Implications

Attempts to understand a role of the P2X₇ receptor in neural or non-neural cells have been hampered by the need to explain why cells would possess such a lethal channel, but there is a considerable call for such action in the neonatal retina. A high number of ganglion cells is lost from the neonatal rat retina in the first two weeks of life, ⁵¹ and the P2X₇ channel may provide a mechanism to thin out unnecessary or unconnected neurons. However, culling surplus cells during neural development cannot be the only role of the P2X₇ receptor in retinal ganglion cells, as the receptor is expressed at a high density in both the neonatal and adult rat retina. ³⁵  

The contribution of P2X₇ receptors to the function of the adult retina depends on both the concentration and the source of stimulation. Although high levels of agonist can be lethal, it is likely that the channel normally functions in a benign mode by increasing the level of Ca²⁺. The ability of NAD⁺ to act as a coagonist for the receptor may help activate the receptor exposed to moderate levels of ATP. ⁵² Several sources of ATP in the retina have been identified, with release from both neural and pigment epithelial tissues. ^{26 53} Of particular interest is the observation that physical stimulation can trigger a release from glial cells adjacent to the ganglion cells. ^{54 55} Although such feedback may be beneficial under control conditions, this pressure-dependent release of ATP suggests that the P2X₇ receptor may contribute to the pathologic loss of ganglion cells in glaucoma. Increased pressure is thought to be a wide-spread trigger of ATP release, ⁵⁶ and our preliminary results suggest that this response also occurs in the retina. Under the sustained period of elevated pressure found in chronic glaucoma, sufficient ATP may be released to overwhelm ecto-ATPases, with an EC₅₀ of 10 to 75 μM ^{57 58} and stimulate the P2X₇ receptor, with an EC₅₀ of 115 to 600 μM. ^{18 59} The continuous release found with sustained elevations of pressure would provide a constant supply of fresh ATP to stimulate the P2X₇ receptor and may lead to the loss of ganglion cells characteristic in glaucoma. Changes in levels of ectoATPases or in expression of the P2X₇ receptor in response to increased pressure could also contribute to the pathologic role of excess ATP. Although the precise conditions that activate the P2X₇ receptor in retinal ganglion cells remain to be determined, the results reported herein clearly indicate that that receptors can lead to an influx of Ca²⁺ and kill the cells. This finding suggests that excessive stimulation of the P2X₇ receptor could contribute to the death of ganglion cells during retinal development and in certain optic neuropathies. Whether stimulation of the P2X₇ receptor can kill neurons elsewhere in the central nervous system may depend on the local arrangement of channels. 

 Supported by National Eye Institute Grants EY-13434 and EY-015537 (CHM), EY-10009 (AML), and Core Grant EY01583, by a University Research Foundation award from the University of Pennsylvania (CHM), by the Jody Sack Fund (XZ, MZ, AML), by a grant from Research to Prevent Blindness (AML), and by Grants 30471850 from the Chinese National Scientific Research Fund and 04009334 from the Guangdong Scientific Research Fund (XZ).

 Submitted for publication January 14, 2005; revised February 4, 2005; accepted February 8, 2005.

 Disclosure: X. Zhang, None; M. Zhang, None; A.M. Laties, None; C.H. Mitchell, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

BzATP-induced elevation of Ca²⁺. (A) In the presence of physiologic levels of divalent cations, a 2-minute application the P2X₇ agonist BzATP (50 μM) led to a rapid, sustained increase in cell Ca²⁺ from retinal ganglion cells loaded with fura-2. The response was reversible and repeatable. (B) The removal of extracellular Mg²⁺ from the solution perfusing the ganglion cells increased the response to 50 μM BzATP. Mg²⁺ was absent throughout the trace. (C) At a concentration of 100 μM, ATP produced a smaller Ca²⁺ response than 10 μM BzATP. The initial transient response to ATP probably represents release from intracellular stores after stimulation of P2Y receptors and was found in approximately half the applications of 100 μM ATP. An enhanced response to BzATP was found in all trials. Horizontal bars: duration of agonist application.

Figure 2.

 

The response to BzATP depended on extracellular Ca²⁺. (A) Retinal ganglion cells loaded with fura-2 were exposed to recurring 15-second applications of 50 μM BzATP followed by a 6-minute wash. This protocol enabled repetitive elevations in Ca²⁺ to be recorded. Similar consecutive increases in Ca²⁺ in response to a 15-second application of BzATP were observed in three trials. All experiments shown were performed in Mg²⁺-free solution. (B) When pulses of 50 μM BzATP were given in the absence of extracellular Ca²⁺, no increase in cellular Ca²⁺ was detected. An extended wash in Ca²⁺-free solution ensured that all BzATP was removed and receptor stimulation ended before Ca²⁺ was reintroduced to the bath. Similar results, with repeatable and reversible responses, were observed in three trials. (C) The antagonist BBG blocked the BzATP-induced increase in cell Ca²⁺. Whereas a 15-second application of 50 μM BzATP led to a clear elevation in control solution, this response was blocked by 1 μM BBG. A small elevation in baseline Ca²⁺ levels accompanied application of BBG. A second response to BzATP was not detected after removal of BBG. Similar responses were found with 100 nM and 10 μM BBG. Horizontal bars: duration of agonist application.

Figure 3.

 

BzATP killed retinal ganglion cells. (A) Incubation with 50 μM BzATP killed retinal ganglion cells in a time-dependent manner. % RGC, the percentage of fluorescently labeled retinal ganglion cells in the presence of BzATP versus the number in parallel experiments in the absence of BzATP measured at the same time point. (•) The mean ± SE (n = 3 at all time points, with each n being the mean cell counts from 80 fields). (B) The number of fluorescently labeled ganglion cells in a mixed retinal culture was decreased in a concentration-dependent fashion after a 24-hour incubation with BzATP. Each point is the mean ± SE (n = 79, 60, 7, and 8 for 0, 50, 100, and 500 μM BzATP, respectively). Solid line: a fit of the function y = y ₀ + ae ^{( −bx )} to the data using the least-squares method, with y ₀ = 55.5, a = 44.3, and b = 0.021, and gives an EC₅₀ of 35 μM. (C) The P2X₇ antagonist oxidized ATP (oATP, 100 μM) prevented the RGC death triggered by 50 μM BzATP. Cells were preincubated with oATP for 2 hours, followed by a 24-hour co-incubation of 100 μM oATP and 50 μM BzATP (n = 24 for each). Central line: the median; dotted line: the mean; top and bottom boundaries: the 25th and 75th percentiles, respectively; top and bottom whiskers: 90th and 10th percentiles, respectively. Data are normalized to the mean control for each day. BzATP was significantly different from the control (P = 0.04) and oATP (P < 0.04 for both); oATP versus control, NS. (D) The nonspecific P2Y and P2X antagonist suramin (30 μM) did not prevent the cell death triggered by 50 μM BzATP (n = 8 for each; BzATP was not different from BzATP+suramin, P = 0.237). Suramin has a minimal effect on P2X₇ receptors at this concentration. Control versus BzATP and versus suramin (P < 0.001 for both). (E) The P2X₇ antagonist BBG reduced the cell death that followed a 24-hour incubation with 50 μM BzATP at concentrations of 100 nM, 1 μM, and 10 μM. BBG was added to the cells 30 minutes before BzATP. (BzATP was significantly different from all other groups; P < 0.05; n = 19 for each.)

Figure 4.

 

Mechanisms of BzATP-triggered cell death. (A) Ganglion cells purified by panning exposed to 50 μM BzATP for 24 hours stained positive for activated caspase with the CaspACE FITC-VAD-FMK stain. Top: Ganglion cells labeled with aminostilbamidine (yellow-green). Middle: staining for activated caspase with the marker CaspACE FITC-VAD-FMK was detected in a subset of ganglion cells pseudocolored red. Bottom: Pseudocolor image showing the overlap of aminostilbamidine (green) and CaspACE FITC-VAD-FMK (red) in three cells (yellow-orange). (B) Retinal ganglion cells did not show uptake of Yo-Pro-1 when observed 25 minutes after application of 50 μM BzATP to mixed retinal cultures. Top: ganglion cells identified by labeling with aminostilbamidine (yellow-green). Middle: uptake of Yo-Pro-1 was observed in numerous smaller unidentified retinal cells (pseudocolored red) but overlap with aminostilbamidine staining was rarely found. Bottom: pseudocolor image showing that aminostilbamidine (green) and Yo-Pro-1 (red) label different cells. (C) The cell death triggered by 50 μM BzATP was attenuated by the L-type Ca²⁺ channel inhibitor nifedipine (50 μM, n = 11, all groups different; P < 0.02). % RCGs and box plots are as described in Figure 3 . Scale bars, 30 μm.

The authors thank Kristine Quinto and Frank Schuettauf for initial technical advice, Andrew Hartwick and William Baldridge for assistance with cell panning, Annmarie Surprenant for advice with the pharmacology, and Tatyana Shekhterman for technical assistance.