Pseudoapoptosis Induced by Brief Activation of ATP-gated P2X7 Receptors*

P2X7 receptors are ATP-gated ion channels primarily expressed on antigen-presenting immune cells where they play a role in the acute inflammatory response. These ion channels couple not only to influx of cations, including calcium, but also to rapid alterations in cell morphology (membrane blebbing, phosphatidylserine exposure, microvesicle shedding). These features resemble the extranuclear events associated with end stages of apoptosis but cell death does not occur if receptor activation is brief. Here we delineate two signaling pathways underlying these apoptotic-like processes. Loss of membrane asymmetry occurs within seconds, which directly triggers cytoskeletal disruption and zeiotic membrane blebbing; this is readily reversible and requires both calcium influx through P2X7 channels and mitochondrial calcium increase but is not associated with cytochrome c release. A slower, calcium-independent, ROCK-1-dependent cascade that does not involve rapid loss of membrane asymmetry but is associated with cytochrome c release is secondarily activated. The ROCK-1 pathway appears largely responsible for cell death, which occurs after prolonged stimulation of P2X7 receptors. We suggest that the former mechanism underlies the reversible pseudoapoptotic events induced by brief activation of P2X7 receptors.

P2X 7 receptors are ATP-gated ion channels primarily expressed on antigen-presenting immune cells where they play a role in the acute inflammatory response. These ion channels couple not only to influx of cations, including calcium, but also to rapid alterations in cell morphology (membrane blebbing, phosphatidylserine exposure, microvesicle shedding). These features resemble the extranuclear events associated with end stages of apoptosis but cell death does not occur if receptor activation is brief. Here we delineate two signaling pathways underlying these apoptotic-like processes. Loss of membrane asymmetry occurs within seconds, which directly triggers cytoskeletal disruption and zeiotic membrane blebbing; this is readily reversible and requires both calcium influx through P2X 7 channels and mitochondrial calcium increase but is not associated with cytochrome c release. A slower, calcium-independent, ROCK-1-dependent cascade that does not involve rapid loss of membrane asymmetry but is associated with cytochrome c release is secondarily activated. The ROCK-1 pathway appears largely responsible for cell death, which occurs after prolonged stimulation of P2X 7 receptors. We suggest that the former mechanism underlies the reversible pseudoapoptotic events induced by brief activation of P2X 7 receptors.
In antigen-presenting cells of the immune system, prolonged application of high concentrations of extracellular ATP leads to apoptotic and/or necrotic cell death via activation of the ATP-gated ionotropic P2X 7 receptor (1-3). P2X 7 receptor coupling has also been associated with classical apoptotic signaling cascades such as activation of stressactivated kinases/c-Jun N-terminal kinases and caspase-3 (4 -7). However, P2X 7 receptors in the immune system can also couple to responses distinct from cell death including cell fusion, cell proliferation, release of pro-inflammatory cytokines, and bone formation (8 -11). Indeed, brief activation of P2X 7 receptors, on the order of seconds to minutes, causes rapid morphological changes, which appear to mimic the extranuclear hallmarks of apoptosis, in particular membrane blebbing, phosphatidylserine exposure (PS flip), 2 and microvesicle shedding. Surprisingly, sub-sequent cell death does not occur (3,10). Is this a type of pseudoapoptosis? That is, might this receptor couple to the same biochemical and cell biology signaling cascades that are now associated with apoptosis (12)(13)(14)(15)) and yet not lead to cell death (16)? Or, might a mechanism distinct from the "classical" apoptotic cytochrome c/caspase3/RhoAassociated kinase (ROCK-1) signaling cascade (17,18) underlie the cellular events associated with brief receptor activation? Elucidation of signaling mechanisms underlying the rapid sequelae of cellular events induced by P2X 7 receptor activation is not only an important aim in the area of ion channel function in immune cells but may also provide significant insight into mechanisms underlying what is regarded as "classical" apoptosis (19 -21). We have addressed this issue by studying the kinetics of cell morphology and mitochondrial alterations in response to brief activation of P2X 7 receptors heterologously expressed in HEK293 cells. We have focused on alterations in cytoskeletal protein elements, changes in lipid asymmetry and mitochondrial function, and the dynamics of membrane blebbing because these are some of the most intensively studied extranuclear events associated with active cell death (i.e. apoptosis) (22)(23)(24)(25)(26).
We find that plasma membrane PS flip and gross mitochondrial swelling with collapse of mitochondrial membrane potential both occur in less than 5-10 s with actin filament/microtubule network disruption and synchronous membrane blebbing occurring some 30 -60 s later. All of these events are fully reversible and not associated with cytochrome c release when receptor activation is less than 20 -30 min. However, longer activation of this receptor leads to release of cytochrome c into the cytosol, which is associated with inevitable cell death. Thus, it appears thatP2X 7 receptorscoupletoatleasttwodistinctmechanisms:acalciumdependent, PS flip-initiated pathway resulting in reversible cytoskeletal disruptions without subsequent cell death and a calcium-independent, ROCK-1-dependent pathway that may be primarily responsible for cell death following prolonged P2X 7 receptor activation.
Microscopy-For transmission electron microscopy, cells were fixed (30 min) in suspension in 4% glutaraldehyde in 0.1 M PIPES buffer at pH 7.2 with 200 ml of 33% hydrogen peroxide per 10 ml of fixative. The cell suspension was centrifuged, the pellet was fixed for a further 30 min, and then washed in 0.1 M PIPES buffer for 1 h. Cells were postfixed in 1% osmium tetroxide in PIPES buffer for 1 h then washed in 0.05 M maleate buffer over 1 h. The preparation was then bulk-stained in 1% uranyl acetate in maleate buffer for 12 h then dehydrated in graded alcohols, infiltrated in Spurr resin, and sectioned at 60 nm on a Riechert ultramicrotome. The sections were examined using a Philips CM 100 electron microscope. Standard video microscopy was performed using a Zeiss Axiovert 100 with Fluar ϫ40 or ϫ100 objective, and a JVC TK-C1380 color video camera coupled with a VHS video recorder. Images were converted from analogue to digital images using the Formac Studio and ProTV software (Formac Electronics, Germany). Image J software (NIH) was used to time-compress videos, which were then saved as Quicktime Movies.
To assess mitochondrial morphology, cells were loaded with 2 M Mitotracker Green for 2-5 min at room temperature. Fluorescent images were collected using either a cooled CCD camera (IMAGO) or a laser scanning confocal microscope (Molecular Dynamics CLSM2010). In fluorescent imaging experiments, images were obtained using an oil immersion ϫ100 Fluar objective by a cooled CCD camera at 0.5-2 Hz, NBD labeling of plasma membrane was as described previously (10). A scanning monochromator selected excitation wavelengths under computer control. Digital images were analyzed using TILLvisION imaging software (Photonics, Planegg, Germany). Excitation/emission wavelengths were 480/488 nm for BTC-AM, 340/380 nm for Fura-2, 560/510 for JC-1, 500 for Rhod-2, and 560 nm for rhodamine phalloidin. Agonists were applied via a gravity-driven superfusion system. Confocal fluorescent imaging of mitochondria (Mitotracker Green) was carried out using the excitation wavelength 488 nm and emission collected via the 530-nm band pass filter (green channel). Images were acquired using an inverted microscope using a ϫ63 oil immersion objective. All experiments were conducted at ambient room temperature 22-24°C. Data are mean Ϯ S.E.
Immunohistochemistry-Cells were washed in PBS and fixed in Zamboni's Fixative (20 min), blocked in 5% goat serum in PBS containing 2% Triton X-100 (30 min), and then incubated with the primary antibody (mouse anti-human vinculin or anti-human ␣-tubulin) for 1-2 h at room temperature or overnight at 4°C. Cells were washed and incubated with the secondary antibody (fluorescein isothiocyanate (FITC)conjugated goat anti-mouse immunoglobulin (IgG) at 1:100 dilution) for 1 h at room temperature. For actin distribution, fixed cells were labeled with rhodamine phalloidin (filamentous actin/F-actin) or DNAse1 (monomeric actin/G-actin) either separately, together, or following labeling for vinculin. Cells were washed in PBS, fixed with Zambonis (10 min), blocked with 1% bovine serum albumin in PBS containing 2% Triton X-100 (20 min), and incubated with rhodamine phalloidin (5 units ml Ϫ1 ) or FITC-DNAse1 for 1 h. For simultaneous labeling of vinculin and F-actin, cells were incubated with rhodamine phalloidin (30 min) following labeling with secondary antibody.
Analysis of Bleb Kinetics and Dimensions-Still images of blebbing cells were used for measurements. Because images were in two dimensions, we estimated the area of each bleb and of the cell itself by approximating each to a circle (drawing a circular region of interest around each) and calculating the area (r 2 ). For each image the sum of the areas of all blebs was expressed as a percent of the area of the cell. This protocol adjusted for differences in cell size and focal plane during video acquisition. Time to onset of membrane blebbing was taken as time when first fully protruding bleb was observed and is accurate to Ϯ 4 s. All measurements were performed blind in terms of knowledge of the experimental condition applied. Curves following either onset of blebbing or extent of blebbing were fit by a least squares method using Kaleidagraph software (Synergy Software) in the form: where Y is response, t is time, V 50 is half-maximal response, and m is slope of the line. The reversal of blebbing was fit only by a smoothing curve using Kaleidagraph.
Cytochrome c Release-To determine cytochrome c release from mitochondria, cytosolic fractions were isolated as described (27). Briefly, 4 ϫ 10 6 cells per sample, were exposed to 100 M BzATP for various times or left untreated. The cells were then lysed under vigorous vortexing in PBS plus 500 mM sucrose containing protease inhibitors (Complete-TM, Roche Applied Science) and further incubated at room temperature for 60 s. Cytosolic fractions were quickly isolated and removed from organelles and cell debris by centrifugation at 14,000 ϫ g for 5 min at 4°C. As reference for total cytochrome c content of the cells, organelle-comprising fractions of untreated cells were lysed in 20 mM dodecylmaltoside in PBS plus protease inhibitors for 2 h under vigorous shaking. The presence of the detergent allowed solubilization of membrane proteins that were then separated from cell debris by centrifugation at 14,000 ϫ g for 5 min at 4°C. Proteins were visualized by Western blotting: SDS sample buffer was added, and proteins were denatured at 100°C for 10 min. They were subsequently separated by SDS-PAGE (15%) and electrophoretically transferred to polyvinylidene difluoride (Immune-blot PVDF, Bio-Rad) membranes. PVDF membranes were rinsed in 10 mM Tris, 250 mM NaCl, pH 7.5, then blocked in 1% bovine serum albumin in TTBS: 10 mM Tris, 250 mM NaCl, 0,1% Tween 20, pH 7.5, for 1 h at room temperature. After a wash with TTBS, a monoclonal anti-cytochrome c Ab (2 g/ml, BD PharMingen clone 7H8.2C12) was added, and the membrane was incubated overnight at 4°C. The membrane was washed twice in TTBS, and a 1:3000 dilution of goat antimouse IgG (HϩL)-AP-conjugated Ab (Bio-Rad) was added in blocking solution. After 2 h of incubation at room temperature the membrane was washed three times and Immun-star substrate from Bio-Rad was applied for 5 min in the dark. Protein bands were revealed by standard photographic procedures. Densitometric semiquantitation of Western blot bands was obtained using the Gene Genius image system from Syngene. Images were acquired and analyzed from films using GeneSnap and Genetools software (Syngene). The reported data represent the ratio between the calculated density of each sample and a control untreated sample present on the same film ( Fig. 7D) or relative to ␤-actin (Fig. 7, B and C) resolved on the same gel using rabbit ␤-actin polyclonal Ab (Abcam, Cambridge, UK).

P2X 7 Receptor-induced Membrane Blebbing and Cytoskeletal
Rearrangements-HEK293 cells stably expressing the rat P2X 7 receptor were used because previous electrophysiological and immunohistochemical studies have shown that virtually all cells exhibit uniform and approximately equal receptor density at the plasma membrane (10,28). The ATP analogue, BzATP, was used in most experiments, because it does not activate the endogenous metabotropic purinergic (P2Y) receptors in these cells (29). The time-to-initial onset of blebbing decreased with increasing agonist concentrations, from 48 Ϯ 3 s (n ϭ 144) at 10 M to 23 Ϯ 1 s (n ϭ 131) at 200 M. As well, the proportion of cells undergoing blebbing was similarly concentration-dependent with 10 M BzATP inducing blebs in Ͻ2% of cells whereas Ͼ97% of cells blebbed in response to 100 M BzATP (Fig. 1, A and B). When these parameters were plotted as a function of the agonist concentration, they were well fit by standard sigmoidal dose-response equations ( Fig. 1, C and D); half-maximal EC 50 values for time-to-onset and percent total cells blebbed were 33 Ϯ 2 M and 72 Ϯ 3 M, respectively. These values are within the range of EC 50 values (3-40 M) for membrane current, dye uptake or calcium entry in these cells under identical conditions (10,28,30) and thus indicate that the signaling pathway(s) underlying these rapid morphological changes are engaged over the same range as that for activation of the P2X 7 ion channel.
Membrane blebs readily reversed within 5-15 min in all cells when BzATP (30, 100 M) was washed out following applications of 1-10 min (Fig. 1, A and E and supplemental Movie 1). Upon washout of agonist, blebs were observed to resorb at a rate corresponding to the total level of blebbing that occurred; that is, smaller blebs reversed more quickly than larger blebs (Fig. 1E). Blebbing was a dynamic process even during washout of agonist, as new blebs appeared while others were resorbed until complete retraction occurred (supplemental Movie 1). This type of dynamic blebbing is termed "zeiotic" and has been associated with apoptotic but not necrotic cell death (31)(32)(33)(34).
Electron microscopy showed that P2X 7 receptor-induced blebs exhibited the same characteristics as classical apoptotic blebs (31,33): they contained only amorphous cytosolic material and no organelles; the double nuclear membrane remained intact during activation for up to 10 min (longest time examined), although the chromatin appeared slightly more condensed when compared with unactivated cells ( Fig. 2A).
Because the actin cytoskeleton is the major factor dictating cell shape and movement, we studied the distribution of polymerized (F-actin) and monomeric (G-actin) actin after 2-10 min exposure to BzATP (Fig.  2). In unactivated cells, a dense network of F-actin fibers traversed the periphery of the cell directly under/at the plasma membrane whereas monomeric G-actin was diffusely localized throughout the cytosol (Fig.  2B, upper panel). One minute after exposure to BzATP (30 M), membrane blebs were rich in monomeric G-actin but relatively devoid of F-actin (n ϭ 8; Fig. 2B, lower panel).
F-actin stress fibers are complexed to integrin receptors via focal adhesion sites, a complex of actin-binding proteins, and signal transduction elements such as tyrosine kinases (35). Disassembly of focal adhesion complexes may account for formation of large blebs devoid of organelles caused by sudden loss of the membrane-cytoskeletal interaction (35,36). Accordingly, we examined the localization of the focal adhesion protein vinculin; an actin-binding protein, which forms part of the focal adhesion complex (37). In control cells, vinculin was distributed at perinuclear regions and at punctate focal complexes at the termini of peripheral F-actin fibers (Fig. 2C, upper panel), but after P2X 7 receptor activation for 2 min, vinculin had accumulated at the cell periphery and in the blebs (Fig. 2C, lower panel) Finally, because the microtubule network regulates contractile and propulsive actions of the actin cytoskeleton, with extensive cross-talk between these two structural elements particularly at the level of focal adhesions (35, 37), we also examined ␣-tubulin distribution after P2X 7 receptor-induced blebbing. In unactivated cells, microtubule filaments extended throughout the cell and terminated within lamellipodia (Fig. 2D, left panel), after 2-min BzATP stimulation, ␣-tubulin was seen to cluster at the leading edge of the bleb (Fig. 2D, middle panel). Distribution of ␣-tubulin had returned to normal within 20 min of agonist removal (Fig. 2D, right panel), by which time all blebs had retracted.
A Calcium Influx Pathway and a Calcium-independent ROCK-1 Pathway Account for Two Types of Membrane Blebbing-Despite these dramatic cytoskeletal rearrangements, neither membrane blebbing nor its reversal were significantly altered by several agents that interfere with cytoskeletal proteins. Specifically, the following treatments were without effect on blebbing or reversibility of blebbing: cytochalasin D (n ϭ 10), which disrupts the actin cytoskeleton by preventing F-actin polymerization, the myosin light chain kinase inhibitor ML-9 (n ϭ 6), or the microtubule depolymerizer, nocodazole (n ϭ 6; Fig. 3A). Because MAPK and Rho-dependent pathways are well known to be intimately involved with cytoskeletal function, we examined the following inhibitors of these pathways but did not observe any significant alterations in the kinetics or extent of P2X 7 receptor-induced membrane blebbing: dominant negative and positive mutations of RhoA (N19Rho and L61Rho), the C3 exoenzyme Rho inhibitor, the p38 MAPK inhibitor SB203580, the p42/44 MAPK inhibitor PD98059, and LY294002, which is a relatively specific inhibitor of phosphatidylinositol 3-kinase at the concentrations used in this study (n ϭ 3-9 for all compounds, final concentrations of compounds used are listed under "Material and Methods, " Fig. 3A). In contrast, the p160 ROCK-1 inhibitor, Y-27632, significantly increased the time to onset of membrane blebbing (Fig. 3A) but had no significant effect on the extent or type of membrane blebbing that occurred. More strikingly, removal of extracellular calcium (with 1 mM EGTA to ensure maximum calcium reduction), clearly altered the features of the blebs as well as increasing the time to onset of initial blebbing (Fig. 3B). The multiple blebbing around the cell circumference seen in control solutions ( Fig. 1A and supplemental Movie 1) was replaced by many fewer (usually only 1 or 2) and much larger blebs, whose neck diameters were up to 10 m (Fig. 3B and supplemental Movie 2). In contrast to the zeiotic nature of the blebbing observed in the presence of calcium, features of the calcium-independent blebs resembled those previously ascribed to necrotic blebs (33). Additional pretreatment with BAPTA-AM to prevent changes in intracellular calcium caused a further reduction in numbers of blebs/cell with concomitant enlargement of individual blebs (Fig. 3B). Under these calcium-free conditions, pretreatment with Y-27632 now led to complete cessation of membrane blebbing (Fig. 3A).
Calcium-independent Blebbing Occurs without Prior PS Flip-We have previously observed that, under normal conditions, PS flip precedes P2X 7 R-induced membrane blebbing, and, for brief agonist applications (Ͻ10 min), reverses within hours without concomitant cell death (10). We have extended these studies by examining more quanti-FIGURE 2. Rapid disruption of cytoskeletal proteins during P2X 7 R activation. A, electron micrographs from control HEK cell and after a 2-min application of BzATP. The bleb is devoid of organelles, mitochondria cisternae are swollen but nuclear membrane is intact. B, distribution of F-actin (red, appears yellow because of overlap with G-actin) and G-actin (green) before and at 2-min stimulation with BzATP (upper and lower panels, respectively); note absence of F-actin in blebs. C, distribution of F-actin (red) and vinculin (green) before and after BzATP stimulation (upper and lower panels, respectively); vinculin is found in blebs. D, distribution of ␣-tubulin before, during, and after wash of BzATP as indicated; ␣-tubulin can be seen to accumulate at the leading edge of blebs with a return to normal distribution within 20 min of wash. Scale bars are 10 m. tatively the kinetics of P2X 7 R-induced PS flip. We also asked whether PS flip preceded and/or accompanied the calcium-independent membrane blebbing, and thus whether it may be an essential upstream signal for initiation of blebbing per se. We used annexin-V-FITC fluorescence to monitor PS flip; however, annexin V binds specifically to PS only in the presence of calcium, and annexin V binding itself blocks further PS translocation (10,38,39). Therefore, we carried out parallel experiments to measure separately at each time point membrane blebbing, annexin V-FITC fluorescence in the presence, or upon reintroduction, of calcium, as well as cell survival 12-h poststimulation (Fig. 4). In control solutions, virtually all cells showed annexin V labeling within 30 s of stimulation with BzATP; membrane blebbing lagged annexin V binding by some 30 -60 s (Fig. 4A). All blebs had retracted within 20 min following a 2-min application of BzATP (Fig. 4A). PS flip was reversed by 3 h following brief BzATP application because no annexin V binding was observed at this time (Fig. 4A). Moreover, no loss of cell number or viability was observed 12 h following the 2-min stimulation with BzATP; indeed, a significant increase in cell number occurred (Fig. 4B). This increased cell proliferation following brief (2 min) stimulation with BzATP occurred only in P2X 7 -expressing cells but not in untransfected or P2X 4 -transfected HEK cells (data not shown). In contrast, similar experiments performed in calcium-free solution on cells preloaded with BAPTA-AM, showed a delay to initial onset of blebbing comparable to that observed in experiments presented in Fig. 3, but no significant annexin V binding was observed at any time point after calcium was reintroduced (Fig. 4A). Interestingly, retraction of these blebs was much delayed; minimal bleb retraction had occurred 20 min after wash into calcium-containing solution although virtually all blebs had retracted within 3 h of wash (Fig. 4A). Subsequent cell death was also not observed under these circumstances although the increased rate of cell proliferation observed in normal solution was not seen (Fig. 4B). We consistently observed membrane blebbing without associated PS flip only when both removal of external calcium and chelation of intracellular calcium with BAPTA-AM loading were carried out (n ϭ 9). When either of these manipulations were carried out separately, membrane blebbing was consistently observed (as described above) but annexin V binding was highly variable and inconsistent (n ϭ 6). As expected (1, 3, 10), prolonged application of BzATP (40 min with calcium) led to death of all cells within 12 h (Fig. 4B). The cell death subsequent to prolonged BzATP application was significantly, but not completely, abrogated by incubation with zVAD-FMK (Fig. 4B). It is most likely that the zVAD-FMK-insensitive cell death following prolonged P2X 7 receptor activation results from necrotic, rather than apoptotic, cell death, and is in keeping with previous studies showing that prolonged activation of P2X 7 receptors can be associated with either or both apoptotic and necrotic cell death (1)(2)(3). Rapid and Reversible Mitochondrial Changes: Depolarization and Swelling without Cytochrome c Release-Electron microscopic examination of unstimulated HEK293 cells showed typical mitochondrial morphology with clear double membranes and internal cisternae (Fig.  5A); dramatic global mitochondrial swelling with loss of cisternae were observed within 2 min of P2X 7 receptor activation (Fig. 5B). We also used confocal microscopy to follow the kinetics of these mitochondrial changes by labeling cells with Mitotracker Green, which accumulates in mitochondria independent of mitochondrial membrane potential, and used this to monitor changes in mitochondrial mass. In resting cells, Mitotracker Green labeled multiple mitochondria ϳ1-m wide, extending as long filaments throughout the cytoplasm (Fig. 5C). Mitochondria began to round up and swell within 5-15 s of BzATP applica-  tion (Fig. 5C). Global mitochondrial swelling of this extent has been considered a trigger for cell death and a terminal event in several cell types (24,25,40,41), but we found that 1-3 h after brief (2 min) activation of P2X 7 receptors, the majority of mitochondria were no longer swollen and had recovered their original filamentous morphology (Fig.  5C). Changes in intramitochondrial free calcium and mitochondrial membrane potential (⌿ m ) during P2X 7 receptor activation were measured using Rhod-2 and JC-1 dyes, respectively (Fig. 6A). A rapid increase in Rhod-2 fluorescence occurred within 5 s of receptor activation, whereas a decrease in JC-1 fluorescence began within 5-10 s (Fig. 6A). Thus, mitochondrial depolarization, increased intramitochondrial calcium, and mitochondrial swelling occur in Ͻ10 s of maximal receptor activation. None of these mitochondrial effects were altered by inhibition of the mitochondrial permeability transition pore using cyclosporin A (1 M pretreatment at 37°C for 1 h, n ϭ 7). The caspase inhibitor zVAD-FMK (100 M for 1 h at 37°C) delayed the onset of mitochondrial disruption (from Ͻ10 s in control to 24 Ϯ 4 s after zVAD-FMK, n ϭ 5), but no other obvious inhibition of mitochondrial disruption was noted. Neither cyclosporin A, nor zVAD-FMK, altered the P2X 7 receptor-mediated membrane blebbing (Fig. 3A). However, removal of extracellular calcium completely prevented mitochondrial swelling (Fig. 6C).
Finally, we asked whether cytochrome c release was associated with these rapid cytoskeletal and mitochondrial rearrangements by immunohistochemistry and immunoblotting (Fig. 7). Fig. 7A shows cyto-chrome c immunostaining (green) with a nuclear counterstain (DAPI, blue). In control cells and after 2 min in the presence of BzATP, all staining was confined to mitochondria even though gross mitochondrial swelling was obvious after 2 min of stimulation (Fig. 7A, middle panel inset). However, after 30 min in BzATP cytochrome c staining was now diffuse throughout the cytoplasm, and many of the swollen mitochondria could be seen to be ruptured (Fig. 7A, lower panel, arrowheads  in inset). Ruptured mitochondria were not observed in cells stimulated for only 2 min. Semiquantitative analysis was carried out by Western blot (Fig. 7, B-D). No detectable release of cytochrome c occurred when P2X 7 receptors were stimulated for 1, 2, or 5 min, whereas after 60 min of continuous stimulation, total cellular cytochrome c was now found in the cytosolic fraction.

DISCUSSION
Here we have identified a sequence of events that we refer to as pseudoapoptosis, which is induced by brief activation of P2X 7 receptors by extracellular ATP. Pseudoapoptosis is an appropriate term because an almost complete repertoire of extranuclear events classically associated with apoptosis is observed: mitochondrial depolarization with gross mitochondrial swelling, rapid increase in both cytosolic and mitochondrial calcium, PS flip, actin filament disruption, and zeiotic membrane blebbing. Despite these dramatic mitochondrial and cytoskeletal alterations, subsequent cell death does not occur, and all events are FIGURE 6. Intramitochondrial calcium increase and collapse of mitochondrial membrane potential accompany calcium-dependent mitochondrial swelling. A, mitochondrial calcium levels monitored using Rhod-2 in response to stimulation with BzATP (30 M indicated by bar) from P2X 7 R-expressing and non-transfected (control) cells. B, mitochondrial potential monitored by JC-1 aggregate fluorescence with 488-nm excitation; a high 590-nm/530-nm emission ratio indicates active, polarized mitochondria, and a fall in this ratio reflects decreased mitochondrial potential. C, photomicrographs of cells loaded with Mitotracker Green and stimulated with BzATP beginning at the 0-s time point. Upper panels are in normal solution, and lower panels are from a separate experiment in calcium-free solution. Cell swelling is apparent within 15 s in control solution, but no changes occur in the absence of external calcium. There is some loss of fluorescence over time, which is caused by photobleaching. Scale bar is 3 m. OCTOBER 7, 2005 • VOLUME 280 • NUMBER 40 readily reversible. Missing from this extranuclear pseudoapoptotic repertoire is release of cytochrome c from the swollen mitochondria. Without this pivotal event, the cell exhibits (reversible) pseudoapoptosis; only when cytochrome c release also accompanies these events does cell death occur. How does the cell biology of P2X 7 receptor-induced pseudoapoptosis compare with the more intensively studied classical apoptosis? What might its mechanisms be and might they provide further insight into molecular mechanisms underlying classical apoptosis? Molecular mechanisms underlying membrane blebbing, particularly apoptotic blebbing, are complex and appear to depend on both the cell type and apoptotic stimulus (14,15,25,41). Nevertheless, most recent studies implicate the kinases ROCK-1 and MLCK, as well as the serine threonine kinase p38 MAPK (23,31). In the present study, we used many inhibitors of these pathways and found no effect on any of the cytoskeletal alterations that occurred with brief (2 min) activation of P2X 7 receptors with the exception of an increased time to initial onset of membrane blebbing after treatment with the ROCK-1 inhibitor, Y-27632, although use of the RhoA dominant negative mutation did not result in a similar alteration. Thus, we found no evidence for a primary involvement of these kinases in P2X 7 R-induced blebbing in the presence of normal extracellular calcium. However, when both extracellular calcium influx and intracellular calcium increases were prevented, a non-zeiotic type of blebbing occurred, which was completely blocked by the ROCK-1 inhibitor, Y-27632. An unmasking of this signaling cascade in the absence of calcium might well be expected as the RhoA/ROCK-1 pathway is known to be calcium-independent (35,37,42). Thus, we can conclude that P2X 7 R activation mediates membrane blebbing by (at least) two distinct paths, a calcium-dependent mechanism and a calcium-independent RhoA/ROCK-1 mechanism. These results, and our proposed model (Fig. 8, see below), may also explain apparently conflicting data obtained from recent studies examining P2X 7 R-mediated membrane blebbing, where either removal of extracellular calcium alone or inhibition of ROCK-1 and p38 MAPK in the presence of calcium were sufficient to block membrane blebbing (34,43,44).

P2X 7 Receptors and Pseudoapoptosis
We observed mitochondrial depolarization, intramitochondrial calcium increase, and gross mitochondrial swelling within 5-10 s following P2X 7 R stimulation; all three events were reversible providing receptor stimulation was brief and, as described above, were not associated with subsequent cell death. Release of cytochrome c was not detected at these early times, but only after prolonged agonist applications, which were also correlated with subsequent cell death. Although the rapid kinetics and reversibility of these mitochondrial alterations are surprising, this is not the first example of mitochondrial swelling and loss of mitochondrial membrane potential in the absence of concomitant cell death. Treatment of human osteosarcoma cells with the mitochondrial prot-  A, summary of sequence of events associated with the calcium-dependent and -independent pathways determined in the present study. Solid lines represent events we have shown directly in this study; we do not know whether cytochrome c release is calcium-independent but we have shown it occurs only after prolonged stimulation and therefore have placed this under the slower pathway. B, proposed mechanisms underlying reversible (pseudoapoptosis) and irreversible (apoptosis) actions of P2X 7 receptor activation. Here we propose that rapid loss of lipid asymmetry (PS flip) directly underlies cytoskeletal protein disruption and zeiotic membrane blebbing. We hypothesize that PS flip is induced by extremely high calcium levels at the plasma membrane because of both calcium influx through the P2X 7 ion channel and mitochondrial calcium release. This pathway is associated with mitochondrial swelling but not with cytochrome c release or subsequent cell death. Prolonged stimulation does result in cytochrome c release, which triggers the well known calcium-independent caspase-3/ ROCK-1 signaling cascade and subsequent cell death. Both pathways are likely to interact with each other, and both pathways may be activated simultaneously under different physiological conditions. See text for further discussion.
onophore, carbonyl cyanide m-chlorophenylhydrazone (CCCP), has been shown to induce mitochondrial permeability transition and mitochondrial swelling without cytochrome c release or subsequent cell death (45,46). There is little doubt that release of intramitochondrial cytochrome c into the cytoplasm leads to apoptotic cell death by binding to APAF and procaspase-9 to form the so-called apoptosome, which in turn leads to activation of the caspase cascade (25). On the other hand, there is much debate as to how cytochrome c is released from mitochondria, with at least five distinct models currently proposed (25). Two of these models involve mitochondrial swelling and loss of mitochondrial membrane potential leading to closure of the voltage-dependent anion channel (VDAC) or opening of the permeability transition pore (PTP) although it should be noted that the strongest evidence for these models of cytochrome c release comes from experiments on isolated mitochondria and other organelles (47). Our present results, together with those of Nagley and co-workers (46,47), fairly conclusively rule out either of these two mechanisms as a primary means for mitochondrial release of cytochrome c in living cells. Remaining models involve formation of a Bax or Bax/VDAC channel or a type of lipid-protein pore (25,48).
All P2X receptors are integral plasma membrane protein ion channels activated by extracellular ATP. They are all cation selective with high permeability to calcium ions, but only P2X 7 receptors couple to the downstream events described in this and previous studies. The intracellular C terminus of the P2X 7 receptor is some 100 -200 amino acid residues longer than the other six P2X subunits, and an intact C terminus is required for downstream signaling (but not for ion flux through the channel). Proteomic analysis has identified protein-protein interactions between the P2X 7 receptor and several cytoskeletal and signal transduction proteins (3,49). Taken together with the present results, we propose the following model that may account for the P2X 7 R-induced pseudoapoptosis elicited by brief agonist application and for the cell death that follows prolonged agonist application (Fig. 8). This model may also be relevant to at least some forms of apoptotic and/or necrotic cell death elicited by other stimuli.
We propose that mitochondria may be tethered to P2X 7 receptors by direct or indirect protein-protein interactions with cytoskeletal proteins. Activation of P2X 7 receptors leads to influx of extracellular calcium through the ion channel and rapid mitochondrial calcium increase, resulting in a localized calcium overload at the plasma membrane-mitochondria-cytoskeleton matrix. This leads to PS flip, either by activation of calcium-dependent flippase-like enzymes, or by calciuminduced lipid-protein interaction, perhaps between PS and P2X 7 or one of its interacting proteins. The consequent loss of lipid topology disrupts the normal interaction between cytoskeletal proteins and plasma membrane culminating in actin filament disruption and induction of zeiotic membrane blebbing. Because cytochrome c is still retained within the swollen, depolarized mitochondria, no caspase cascade to cell death occurs, and all events reverse as calcium levels fall and lipidprotein homeostasis is re-established. However, prolonged stimulation leads to development of mitochondrial channels (Bax or Bax/VDAC) or lipid-protein pores, which release cytochrome c and thus initiate apoptosome formation with classical apoptotic ROCK-1-dependent downstream signaling. What controls this shift from cytochrome c retention to release during P2X 7 receptor activation is unknown, but our results suggest it is initiated by a distinct signaling pathway independent of the initial rapid mitochondrial calcium rise and plasma membrane lipidprotein alterations, which we propose as the primary mechanism underlying P2X 7 R-induced pseudoapoptosis. We base this suggestion on our findings that the ROCK-1-dependent blebbing was clearly observed only when both calcium influx and intracellular calcium rises were blocked, and because this calcium-independent blebbing was not associated with PS flip. Thus, in our model, PS flip is the initial upstream effector underlying P2X 7 R-induced pseudoapoptosis whereas it is only a late downstream consequence of the cytochrome c-based apoptotic cascade resulting from prolonged receptor activation.
Apoptotic membrane blebbing has often been distinguished from other types of blebbing by its zeiotic nature; that is, the process of continuous and dynamic protrusions and retractions, in contrast to necrotic membrane blebbing, which is characterized by blebs that do not retract but gradually enlarge to diameters Ͼ10 -20 m (33). In the presence of calcium, activation of P2X 7 receptors induces zeiotic blebbing, albeit on a time scale at least an order of magnitude faster than the currently reported apoptotic zeiosis (e.g. supplemental Movie 1 and Ref. 10). We and others (28,34,43,44) have previously described membrane blebbing in response to activation of P2X 7 receptors heterologously expressed in HEK cells or endogenously expressed in macrophage and other immune cells, and had assumed it was, indeed, an indication of forthcoming cell death. Results from the present study show that this is not the case for brief receptor activation. Whereas the physiological role(s) for the rapid, reversible actin filament rearrangements and subsequent zeosis induced by PS flip in response to brief P2X 7 receptor activation are not known, previous studies have linked activation of this receptor to cell proliferation in human lymphocytes and to cell-cell fusion in other immune cells (8,9,50,51), both processes of which require concerted cytoskeletal reorganization and regulated plasma membrane disruption. The pseudoapoptosis identified in the present study may provide a mechanism underlying P2X 7 R-generated giant cell formation at sites of inflammation or cell proliferation during tumorigenesis.