Involvement of Sodium in Early Phosphatidylserine Exposure and Phospholipid Scrambling Induced by P2X7 Purinoceptor Activation in Thymocytes*

Extracellular ATP (ATPec), a possible effector in thymocyte selection, induces thymocyte death via purinoceptor activation. We show that ATPec induced cell death by apoptosis, rather than lysis, and early phosphatidylserine (PS) exposure and phospholipid scrambling in a limited thymocyte population (35–40%). PS externalization resulted from the activation of the cationic channel P2X7 (formerly P2Z) receptor and was triggered in all thymocyte subsets although to different proportions in each one. Phospholipid movement was dependent on ATPec-induced Ca2+ and/or Na+ influx. At physiological external Na+ concentration, without external Ca2+, PS was exposed in all ATPec-responsive cells. In contrast, without external Na+, physiological external Ca2+ concentration promoted a submaximal response. Altogether these data show that Na+ influx plays a major role in the rapid PS exposure induced by P2X7 receptor activation in thymocytes.

Extracellular ATP (ATP ec ), a possible effector in thymocyte selection, induces thymocyte death via purinoceptor activation. We show that ATP ec induced cell death by apoptosis, rather than lysis, and early phosphatidylserine (PS) exposure and phospholipid scrambling in a limited thymocyte population (35-40%). PS externalization resulted from the activation of the cationic channel P2X7 (formerly P2Z) receptor and was triggered in all thymocyte subsets although to different proportions in each one. Phospholipid movement was dependent on ATP ec -induced Ca 2؉ and/or Na ؉ influx. At physiological external Na ؉ concentration, without external Ca 2؉ , PS was exposed in all ATP ec -responsive cells. In contrast, without external Na ؉ , physiological external Ca 2؉ concentration promoted a submaximal response. Altogether these data show that Na ؉ influx plays a major role in the rapid PS exposure induced by P2X7 receptor activation in thymocytes.
Extracellular ATP (ATP ec ) 1 induces thymocyte apoptosis and may play a role in thymocyte maturation (1)(2)(3)(4). A major change of cell membrane during apoptosis, slightly preceding nuclear condensation (5), is the exposure on the exoplasmic cell surface of phosphatidylserine (PS) (6 -9), a phospholipid normally maintained on the inner membrane leaflet by the aminophospholipid translocase (APLT) (10). This process is associated with a general loss of the asymmetric distribution (scrambling) of all the other major phospholipids (6,7,9). The presence of specific PS receptors on macrophages potentially allows a rapid elimination of apoptotic cells by phagocytosis before their lysis (11). Rapid PS exposure and scrambling are generally observed after platelet activation or treatment of cells with a Ca 2ϩ iono-phore, both conditions promoting a drastic and rapid rise in cytosolic Ca 2ϩ concentration ([Ca 2ϩ ] i ) (7). An involvement of Ca 2ϩ has been also reported in PS exposure during apoptosis (6,7,12). Ca 2ϩ acts by inhibiting APLT and by targeting scramblase (6,7,13) or other molecular complexes, which could contain phosphoinositides (14,15). Recent studies show that the rapid Ca 2ϩ -induced PS exposure is correlated with scramblase expression (16), in contrast to the delayed PS exposure associated with apoptosis (17,18), in agreement with the involvement of other phospholipid transporters in apoptotic pathways.
ATP ec is able to evoke physiological responses in a wide variety of cells and tissues. ATP is concentrated in cell cytosol and in exocytic vesicles, reaching 2-4 and 100 mM, respectively (19), and can be merely released from damaged cells, transported through carriers or channels, or secreted (19 -21). In the extracellular medium, its lifetime is very short due to its rapid degradation by ectonucleotidases and ecto-ATPases (19). ATP ec binds to distinct types of purinergic receptors classified into two subfamilies: metabotropic P2Y receptors coupled to Gprotein (P2Y1-6 and P2Y11) and ligand-gated ion channel P2X receptors (P2X1-7) (22). P2X7 receptor, previously designated as P2Z receptor (23), utilizes extracellular ATP 4Ϫ to increase cationic permeability with consecutive plasma membrane depolarization. Another feature of the P2X7 receptor is its ability to form or activate a non-selective pore, permeable to molecules with molecular mass Ͻ400 Da in thymocytes, when the ATP ec application is prolonged or repeated (24,25). Activation of the P2X7 receptor triggers various physiological processes, including cell death by apoptosis or necrosis (26) or interleukin-1␤ release (22,27,28). PS is rapidly externalized upon P2X7 receptor activation during peripheral T cells apoptosis (29,30) and precedes interleukin-1␤ release from lipopolysaccharideprimed monocytes (28). PS exposure, Ca 2ϩ influx, and pore formation are dependent on the cytoplasmic domain of the P2X7 receptor, which harbors a putative tumor necrosis factor receptor-related death domain and a Src homology 3-binding domain (29).
In this work, we show that ATP ec induced early PS exposure, 1-oleoyl-2-[6-(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]caproylsn-glycero-3-phosphocholine (NBD-PC) scrambling, and apoptosis in 35-40% of the thymocytes. PS exposure and NBD-PC scrambling were dependent on an increase in Ca 2ϩ and/or Na ϩ influx resulting from P2X7 activation. At physiological external concentrations, Na ϩ was more potent than Ca 2ϩ to induce PS exposure. Portions of this work have appeared in a preliminary report (31). * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Recipient of a grant from the Association de la Recherche contre le Cancer (ARC).

Flow Cytometry Analysis
Endogenous PS Externalization-When cells were incubated in Ca 2ϩ -containing medium, PS exposure was assayed from the binding of FITC-annexin-V in the presence of PI (50 g/ml), a marker of plasma membrane integrity used to evaluate the percentage of necrotic cells. PSϩ/PIϪ cell population represents non-necrotic cells with externalized PS. Alternatively, when cells were incubated in a medium devoid of Ca 2ϩ , factor Va, another specific ligand of PS whose binding is Ca 2ϩindependent, was used as described previously (32). Binding of factor Va was detected by indirect immunofluorescence. The fixation of factor Va on the cell surface altered the plasma membrane integrity as reflected by the uptake of PI in all PSϩ cells. Consequently the responsive cells were expressed as total PSϩ cells. However, the proportion of the necrotic cells, estimated in independent assays by measurement of PI uptake, was found to be less than 10% after ATP ec treatment, a proportion similar to that observed in control cells. The percentage of PSϩ cells determined by factor Va binding was always slightly higher than that obtained by FITC-annexin-V staining, suggesting that the affinity of factor Va for PS was higher than that of FITC-annexin-V.
NBD-PC Internalization-The transverse redistribution of the PC analog was assayed as described previously (6,33). Briefly, after incubation of thymocytes for 10 min with 1 mM ATP ec , the fluorescent analog NBD-PC (62.5 nM final concentration representing about 1.5% of cellular phospholipids) was loaded into the cells by 6-min incubation at room temperature. The percentage of cells with internalized PC was assayed after extraction by fatty acid-free bovine serum albumin (2 min at 4°C) of the fraction of the analog present in the external leaflet. Samples were sorted by flow cytometry after addition of PI to evaluate the proportion of necrotic cells.
Plasma Membrane Depolarization-Thymocytes (2⅐10 6 /ml) were stained with 300 nM DiBac4(3) for 20 min at 37°C. Fluorescence intensity of the DiBac4(3) probe increased when cells were depolarized. In these experiments, PI uptake was used to exclude necrotic cells and to determine plasma membrane integrity in non-necrotic cells.
Ethidium or Propidium Uptake-After various times of incubation at 37°C, an aliquot of cells (2⅐10 6 /ml) was stained with PI (50 g/ml) or ethidium bromide (10 g/ml) for 3 min at room temperature. The extent of the pore opening induced by ATP ec was evaluated from the fluorescence of propidium or ethidium bound to nuclear DNA and detected by flow cytometry.
Detection of Surface Markers CD4 and CD8 -The surface markers CD4 and CD8 were detected by flow cytometry after staining thymocytes with R-phycoerythrin-anti-CD4 and either FITC-anti-CD8 or biotin-anti-CD8 and peridin chlorophyll a protein-streptavidin.
Cytosolic Calcium Concentration-Thymocytes (2⅐10 6 /ml) were loaded with 2 M fluo4-AM (15 min at 37°C). PI uptake was used to exclude necrotic cells. The variations in [Ca 2ϩ ] i were determined in non-necrotic cells. [Ca 2ϩ ] i was calculated from the following equation: where K d is the dissociation constant of the Ca 2ϩ -fluo4 complex (345 nM) (34), F max is the maximum fluorescence (obtained after addition of 8 M A23187 and 1.4 mM Ca 2ϩ for 5 min), and F min is the minimum fluorescence (obtained after addition of 8 M A23187 and 5.4 mM EGTA for 5 min). In control cells one fluorescence peak was observed, whereas after ATP ec stimulation a second population with higher fluorescence was apparent (high Ca 2ϩ cells). F is the geometric mean fluorescence value of this high Ca 2ϩ cell population.
In all these experiments, cells were sorted by flow cytometry using a BD Biosciences FaxVantage cytometer. Fluorochrome excitation was performed by argon laser at 488 nm, and fluorescence emissions were analyzed with a 530 nm filter for FITC, NBD-PC, DiBac4(3), and fluo4; with a 575 nm filter for ethidium and R-phycoerythrin; with a 630 nm filter for PI; and with a 660 nm filter for peridin chlorophyll a protein.

Measurement of Calcium Influx
Cells (5⅐10 6 /ml) were loaded with 4 M fura2-AM (30 min at 37°C) and resuspended in a Ca 2ϩ -free RPMI 1640 medium supplemented with 150 M EGTA. Fura2 fluorescence was recorded continuously in a thermostatted and magnetically stirred cuvette of a spectrofluorometer. The fluorescence emission ratio of the probe excited at 340 and 380 nm reflected [Ca 2ϩ ] i .

Measurement of Nuclear Apoptotic Events
Nuclear condensation was quantified by fluorescence microscopy after DNA staining with Hoescht 33342 (10 g/ml) and PI (50 g/ml) for 10 min. Approximately 200 randomly selected cells were scored for each experimental condition. Cells exhibiting blue or red condensed or fragmented nuclei were considered apoptotic. Cells with red nuclei without signs of condensation or fragmentation were considered necrotic.
For detection of DNA fragmentation, cells (3⅐10 6 ) were washed and lysed in 10 mM Tris-HCl, pH 7.4, 0.2% Triton X-100, 1 mM EDTA, 100 g/ml proteinase K for 10 min on ice followed by 60 min at room temperature. After centrifugation (30 min, 13,000 ϫ g, 4°C), DNA in the supernatant was precipitated overnight in cold absolute ethanol containing 0.3 M sodium acetate at Ϫ20°C. After centrifugation (30 min, 13,000 ϫ g, 4°C), the pellet was resuspended in water and incubated for 30 min at 37°C with 50 g/ml DNase-free RNase. DNA concentration was quantified by UV spectrometry at 260 nm. About 5 g of DNA was mixed with gel loading solution and incubated for 5 min at 65°C before electrophoresis in a 1.5% agarose gel containing ethidium bromide. DNA was visualized by UV light.

Data Analysis
Data are means Ϯ S.E. of a minimum of three independent experiments. Differences between means were evaluated by paired t test with p Ͻ 0.05 being taken as the level of significance.

ATP ec Induces Apoptosis and PS Exposure in a Limited
Thymocyte Population-Cell death was measured after 5-h incubation of freshly isolated thymocytes in the presence of ATP ec . Nuclear condensation was significant at 0.5 mM ATP ec in 20 -25% of the cells and maximal at 1 mM affecting 35% of the total population (Fig. 1A). Thymocyte apoptosis was confirmed by visualization of DNA fragmentation on agarose gel electrophoresis. Oligonucleosome formation induced by 1 mM ATP ec was comparable to that promoted by 1 M dexamethasone, a well known inducer of thymocyte apoptosis (Fig. 1B). When compared with spontaneous thymocyte apoptosis, the effect of 1 mM ATP ec on nuclear condensation was significantly increased after 4-h treatment and reached a maximum (35-40%) after 6 h (Fig. 1C). Incubation for up to 9 h did not change significantly the percentage of cells with condensed nuclei and marginally increased the percentage of necrotic cells (11%), indicating that under these experimental conditions ATP ec induced apoptosis preferentially to necrosis in a limited population of thymocytes.
During the apoptotic process, PS exposure was reported to be concomitant or to precede slightly nuclear events such as nuclear condensation (5). ATP ec induced PS exposure and nuclear condensation within the same range of concentrations (0.5-3 mM) with a maximal effect at 1 mM (Figs. 1A and 2A). Surprisingly PS exposure was already significant in 15% of cells after 6-min treatment with 1 mM ATP ec and was maximal after 20 -30 min, affecting about 40% of the thymocytes (Fig. 2B). These results show that PS exposure was a very early event in thymocytes, preceding nuclear alterations. After 5-h incubation with 1 mM ATP ec , both PS exposure and nuclear condensation, observed by light microscopy after FITC-annexin-V binding and Hoescht staining, were detected in the same cell population (data not shown). . DNA fragmentation was determined by agarose gel electrophoresis. 100-base pair DNA ladder was used as molecular marker (M). C, thymocytes were incubated without (control) or with 1 mM ATP ec in 5% FCS/RPMI for various times. The percentage of condensed nuclei was evaluated as in A (means Ϯ S.E. of three to seven independent experiments).

FIG. 2. Effects of ATP ec on PS exposure.
A, thymocytes were incubated at 37°C for 30 min in 5% FCS/RPMI with different concentrations of ATP ec . The percentage of PS-exposing cells impermeable to PI (PSϩ/PIϪ) was evaluated from FITC-annexin-V binding and PI uptake by flow cytometry. B, thymocytes were incubated without (control) or with 1 mM ATP ec in 5% FCS/RPMI for various times, and the percentage of PSϩ/PIϪ was determined as above. Results are means Ϯ S.E. of three to seven independent experiments. C, after incubation for 30 min with 1 mM ATP ec , thymocyte subsets (DN, DP, SPCD4ϩ, and SPCD8ϩ) were characterized by CD4/CD8 staining. In each subset, PS exposure was detected simultaneously by FITC-annexin-V binding by flow cytometry. Results represent the percent distribution of thymocyte subsets and the percentage of PSϩ cells in each subset (mean Ϯ S.E. of three independent experiments).
Differentiation of immature thymocytes into mature T cells takes place in the thymus and proceeds via an ordered sequence of developmental steps characterized notably by variable expression of CD4 and CD8. Early precursor cells are initially double negative CD4ϪCD8Ϫ (DN) and go through a double positive CD4ϩCD8ϩ (DP) intermediate state before differentiating into single positive CD4ϩCD8Ϫ or CD4ϪCD8ϩ (SPCD4ϩ or SPCD8ϩ, respectively). The distribution of the thymocyte population and the percentage of PS-exposing cells in each subset (DN, DP, and SP, either CD4ϩ or CD8ϩ) were analyzed by flow cytometry. As previously reported (2), DP accounted for about 80% of the population (Fig. 2C). ATP ec induced PS externalization in each thymocyte subset. The most mature thymocytes (SPCD4ϩ or SPCD8ϩ) were also those in which the percentage of cells exposing PS was significantly the highest (68 and 73% versus 43 and 29% in DN and DP, respectively) (Fig. 2C). These data show that, as far as PS exposure is concerned, ATP ec responsiveness was dependent on the maturation state of thymocytes.
P2X7 Purinoceptor Is Involved in ATP ec -induced PS Exposure-Thymocytes express different classes of membrane receptors activated by purine ligands (1,3). To identify which purinoceptor(s) was mediating early PS exposure in thymocytes, the effect of various purine agonists and antagonists was investigated. Neither adenosine (a P1 receptor agonist) nor UTP (a P2Y1, P2Y4, and P2Y6 agonist) (22) was able to induce PS exposure. ␣␤mATP, a P2X1 or P2X3 receptor agonist (25), was unable to induce PS exposure (Fig. 3A). In contrast, 0.1 mM BzATP, a P2X7 agonist with a 10-fold higher affinity than ATP ec (22,25), promoted PS externalization to the same extent as did 1 mM ATP ec (Fig. 3A). Furthermore antagonists of P2X (suramin or PPADS) (22,25), at concentrations known to in-hibit P2X1 receptor (1), had no effect on PS exposure induced by 1 mM ATP ec (Fig. 3B), whereas oxidized ATP, a P2X7 antagonist (22,25), significantly inhibited ATP ec -induced PS externalization (Fig. 3B). As P2X7 is activated by ATP 4Ϫ but not by Mg-ATP (22,25), chelation of ATP ec by high Mg 2ϩ concentrations antagonizes P2X7 activation. Pretreatment of thymocytes with 15 mM Mg 2ϩ significantly inhibited PS exposure induced by ATP ec without affecting PS exposure in control cells (Fig.  3B). Furthermore ADP induced PS exposure in a significantly smaller percentage of cells than did ATP ec , whereas 1 mM AMP was without effect (Fig. 3A) as expected from their respective affinities for the P2X7 purinoceptor (2,23,35). PS exposure induced by 1 mM ADP was inhibited by oxidized ATP confirming that this effect was mediated through P2X7 activation (data not shown). PS exposure required ATP ec concentrations above 0.1 mM ( Fig. 2A) in agreement with the range of concentrations reported to activate P2X7, excluding a role for P2X1 purinoceptor, whose activation occurs at concentrations below 0.1 mM (25). Altogether these data provide evidence for the implication of P2X7 in ATP ec -induced PS exposure.
P2X7 Receptor Is Initially in Channel Conformation during PS Exposure-To determine whether P2X7 was in a channel or pore conformation under the conditions leading to early PS exposure, we measured the uptake of two impermeant nucleus markers of different molecular sizes, ethidium (314 Da) and propidium (414 Da), in response to 1 mM ATP ec . Fig. 4A shows that, during the first 20 min of incubation, there was no uptake of propidium, and the uptake of ethidium, observed in less than 15% of cells, was not significantly different with or without ATP ec . Furthermore the level of fluorescence intensity was the same in control or ATP ec -treated cells (data not shown). These data indicate that P2X7 was not in a pore conformation at least for the period during which PS became externalized in thymocytes upon treatment with 1 mM ATP ec . Nevertheless, after 40 min of ATP ec treatment, only ethidium and not propidium uptake was observed in a significant percentage of cells (Fig.  4A), showing that the pore formation was a time-dependent and size-limited process in thymocytes.
P2X7 is an ATP-gated nonspecific cation channel mediating fast permeability changes to Na ϩ , K ϩ , and Ca 2ϩ (22). When thymocytes were incubated in a Ca 2ϩ -free medium, there was no change in [Ca 2ϩ ] i upon addition of 1 mM ATP ec , showing that this treatment did not induce any Ca 2ϩ release from intracellular stores (Fig. 4B). Nevertheless subsequent addition of Ca 2ϩ to the medium triggered a significant increase in [Ca 2ϩ ] i , reflecting Ca 2ϩ influx in ATP ec -treated cells (Fig. 4B). Attempts to measure Na ϩ influx by using Sodium Green-AM or sodiumbinding benzofuran isophthalate-AM (Molecular Probes), two FIG. 4. Effects of ATP ec on PI and ethidium uptake, Ca 2؉ influx, and membrane depolarization. A, thymocytes were incubated without (control) or with 1 mM ATP ec for different times. After incubation, cells were stained with PI (Pr) or ethidium bromide (Et), and the percentage of cells with an increased fluorescence was measured by flow cytometry (means Ϯ S.E. of three independent experiments). B, after loading with fura2-AM, thymocytes were incubated in a Ca 2ϩ -free medium containing 150 M EGTA. The fura2 fluorescence ratio (340/ 380 nm) was used to estimate the variations in [Ca 2ϩ ] i . Gray line, control thymocytes before and after addition of 2 mM Ca 2ϩ ; black line, thymocytes incubated with 1 mM ATP ec before and after addition of 2 mM Ca 2ϩ . C, after loading with DiBac4(3), thymocytes were incubated with 1 mM ATP ec for various times. Histograms of DiBac fluorescence were obtained by flow cytometry. specific probes for Na ϩ detection, were unsuccessful, presumably as a result of inefficient probe loading in thymocytes. However, addition of 1 mM ATP ec induced a rapid depolarization of the plasma membrane detected from the increase in DiBac4(3) fluorescence (Fig. 4C), providing indirect evidence for the occurrence of ATP ec -induced Na ϩ influx (35). Altogether these data confirm that P2X7 in thymocytes is a rapidly gated cation channel.
ATP ec Induces Ca 2ϩ -dependent or Na ϩ -dependent PS Exposure-To determine a possible involvement of ATP ec -induced Ca 2ϩ influx in PS exposure, the percentage of cells with externalized PS (PSϩ cells) was determined after 20-min incubation with 1 mM ATP ec in Na ϩ -free choline media containing different Ca 2ϩ concentrations. Increasing external Ca 2ϩ concentration resulted in a progressive elevation in [Ca 2ϩ ] i (Fig. 5A), detectable as soon as 1 mM external Ca 2ϩ , in a maximum of about 40% of thymocytes (data not shown), named high Ca 2ϩ cells. Similarly the percentage of PSϩ cells induced by ATP ec increased with external Ca 2ϩ , reaching a maximum as soon as 2.5 mM Ca 2ϩ , not significantly different from the percentage detected in 5% FCS/RPMI (Fig. 5A, dotted line). Remarkably, at 0.5 mM Ca 2ϩ in choline medium, PS exposure was observed in only a significantly smaller percentage of cells than that detected in 5% FCS/RPMI, although the latter medium contained about 0.5 mM Ca 2ϩ (Fig. 5A, dotted line). This difference suggests that an additional mechanism, independent of Ca 2ϩ , was involved in PS exposure in 5% FCS/RPMI.
In addition to Ca 2ϩ uptake, P2X7 activation triggers K ϩ efflux and Na ϩ influx (22). Interestingly P2X7 channel can be activated in the absence of external Ca 2ϩ and Na ϩ (36). Thus, in the choline medium devoid of both Ca 2ϩ and Na ϩ , ATP ec could induce K ϩ efflux but was unable to promote significant PS exposure (Fig. 5, A and B), suggesting that K ϩ efflux alone could not be responsible for PS exposure. To study the possible involvement of Na ϩ in ATP ec -induced PS exposure, we determined the percentage of PSϩ cells in a Ca 2ϩ -free choline media containing different Na ϩ concentrations. The percentage of PSϩ cells, induced by 20-min treatment with 1 mM ATP ec , augmented at external Na ϩ concentrations above 70 mM, affecting 38% of cells at 139 mM Na ϩ , a percentage not significantly different from that observed in 5% FCS/RPMI (Fig. 5B, dotted line). Under these conditions, the effect of external Na ϩ on PS exposure can be attributed to ATP ec -mediated Na ϩ influx.
To provide further evidence for a Na ϩ -dependent mechanism of PS exposure, we tested the effect of gramicidin D and monensin, ionophores that induce Na ϩ influx in cells suspended in media containing exclusively Na ϩ . Fig. 5C shows that incubation of thymocytes for 15 min in NaCl medium, but not in choline medium, with 1.5 M gramicidin D induced PS exposure in about 25% of cells. Increasing the time of incubation or the gramicidin D concentration induced thymocyte necrosis as detected by PI uptake (data not shown). In the same way, PS exposure was induced by monensin in the presence of Na ϩ in the medium (Fig. 5C).
ATP ec Induces Ca 2ϩ -dependent or Na ϩ -dependent Phospholipid Scrambling-PS exposure is generally accompanied by the transmembrane redistribution of the other major phospholipids, a process known as scrambling. To investigate whether this scrambling process also occurs in thymocytes activated by ATP ec , PC redistribution was studied using a fluorescent analog of PC (NBD-PC). This method allows the determination of the percentage of cells exhibiting a transbilayer movement of PC at the time of measurement, whereas the factor Va binding assay provides the percentage of cells having exposed PS over the duration of the ATP ec treatment. In choline medium con-taining 5 mM Ca 2ϩ , 10 min after addition of 1 mM ATP ec NBD-PC was internalized in about 35-40% of thymocytes, a percentage similar to that of PSϩ cells under the same conditions (Fig. 6A). Addition of 15 mM MgCl 2 after 10-min incuba- tion with ATP ec to inactivate P2X7 did not affect significantly NBD-PC internalization and endogenous PS exposure (Fig.  6A), indicating that transient activation of P2X7 was sufficient for inducing and maintaining phospholipid scrambling. In fact, under these conditions, [Ca 2ϩ ] i remained elevated as detected by the maintenance of a high fluorescence level of fluo4 (data not shown).
Interestingly treatment of thymocytes for 10 min with 1 mM ATP ec in NaCl medium also induced NBD-PC redistribution (Fig. 6B), showing that Na ϩ influx, as well as Ca 2ϩ influx, was able to mediate ATP ec -induced phospholipid scrambling. When ATP ec after 10-min incubation was chelated by high Mg 2ϩ concentration, NBD-PC scrambling was stopped, and endogenous PS relocated on the inner membrane leaflet (Fig. 6B), showing that, in contrast to the effect of high external Ca 2ϩ oncentration, the Na ϩ effect on scrambling was rapidly reversible. However, under the same conditions, the plasma membrane remained depolarized as detected after DiBac4(3) staining (data not shown), suggesting that membrane depolarization was triggered at a much lower Na ϩ concentration than PS exposure. This hypothesis was supported by the observation that 25 mM external Na ϩ was sufficient to induce a significant depolarization of thymocytes (data not shown) but not PS exposure, which required about 105 mM external Na ϩ (Fig. 5B). DISCUSSION ATP ec has been reported to induce cell death by lysis or apoptosis in a variety of cells, including immune cells, such as dendritic cells, macrophages, and thymocytes (22). Our data show that ATP ec induced a rapid PS exposure in thymocytes, which was already significant after 10 -15 min. Following a longer time of incubation, PS exposure and nuclear condensation occurred in the same cell population, amounting to 35-40% of total thymocytes. Cells exhibiting PS exposure and nuclear condensation remained impermeable to PI, and apoptosis was confirmed by DNA fragmentation observed in cellular extracts. Consequently, under these experimental conditions, thymocyte death induced by ATP ec occurred by apoptosis rather than by lysis. ATP ec -induced apoptosis of thymocytes has been reported to be mediated through activation of P2X1 and P2X7 (1)(2)(3)(4). Our pharmacological studies provide evidence for the involvement of P2X7 in PS exposure. We confirm also that, in thymocytes, ATP ec triggers a significant and sustained Ca 2ϩ influx without releasing intracellular Ca 2ϩ (3,24,35,37), a rapid membrane depolarization (4,24,35), and the progressive formation of a limited size pore permeable to ethidium and not to PI (22,24,35). All these characteristics are hallmarks of P2X7 activation.
PS externalization is known to play a critical role in cell phagocytosis by macrophages (38 -40). Although PS exposure is considered to be a hallmark of apoptosis, it can be dissociated from other features of the apoptotic process, for instance caspase activation or DNA fragmentation, as PS-dependent phagocytic recognition of cells can occur earlier or independently of these latter events (41,42). We have observed that, after a short treatment (30 min) with ATP ec , thymocyte phagocytosis by thioglycollate-elicited peritoneal macrophages was significant and inhibited when external PS was masked by annexin-V binding. 2 Consequently early PS exposure, mediated by ATP ec via P2X7 activation and preceding further steps of apoptosis, could lead to recognition and phagocytosis of thymocytes. Such a possibility for a rapid thymocyte elimination is supported by the observations of Surh and Sprent (43) that apoptotic cells were mainly present within phagocytic cells in the thymus. Furthermore thymocyte elimination, studied in fetal thymic organ culture, suggests an implication of ATP ec and potentially of P2X7 in thymocyte death by neglect (3). In this context, the early P2X7-induced PS exposure could be involved in thymocyte elimination during thymic selection. From the various P2X7-mediated responses (Ca 2ϩ influx (35,37), plasma membrane depolarization (4), or ethidium uptake (35)) within the different subsets of thymocytes or in peripheric T lymphocytes, previous studies have attempted to implicate ATP ec in thymocyte maturation and/or in peripheral T cell regulation. DN subset was the most responsive to P2X7 activation according to Nagy et al. (4), whereas, in other studies, the most mature SP subsets, notably SPCD4ϩ, exhibited the strongest response (35,37). This latter observation was supported by the persistence of the P2X7 response in peripheral T cells (30,35). We show that P2X7 activation induced PS exposure in all subsets and predominantly in the most differentiated SP cells. This characteristic is likely due to a higher level of P2X7 expression in SP subsets in relation to the thymocyte maturation state (35). Furthermore P2X7 activation, induced by ATP ec in splenic T cells (29) or by ADP-ribosylation in lymph node T cells (30), leads to apoptosis with an early PS exposure, suggesting a role of P2X7 in T cell death, for instance to eliminate potentially autoreactive T cells during inflammatory reaction (30). Thus, early PS externalization induced by P2X7 activation could be essential in the rapid elimination of peripheral T cells by phagocytosis.
Our data show that ATP ec induced both early PS exposure and NBD-PC internalization, reflecting phospholipid scrambling, a process generally attributed to an increase in [Ca 2ϩ ] i in cell activation (7) and apoptosis (6,7,12). However, in many instances there was no correlation between scrambling and increased [Ca 2ϩ ] i . In Jurkat T cells, the increase in [Ca 2ϩ ] i induced by Ca 2ϩ ionophore was not sufficient by itself to trigger scrambling, leading to the hypothesis that the ionophore has an additional effect, possibly implicating mitochondria (44). In thymocytes, induction of apoptosis and PS exposure by dexamethasone, gliotoxin, and thapsigargin was not correlated with an increase in [Ca 2ϩ ] i (45). In the same way, during in vitro attack of hepatocytes by natural killer cells, there was no evidence for a straightforward relationship between elevation of [Ca 2ϩ ] i and PS externalization (46). Ca 2ϩ is not the only known inducer of phospholipid scrambling: a decrease in pH stimulated NBD-phospholipid scrambling in inside-out vesicles (13,47), and polyamines were implicated in PS exposure in apoptosis (48). In platelets stimulated with collagen, Na ؉ influx, mediated through activation of the Na ϩ /H ϩ exchanger, was involved in the generation of procoagulant activity (49), a process generally initiated by PS externalization.
We investigated the respective roles of Ca 2ϩ and Na ϩ in ATP ec -induced PS externalization by manipulating the cation concentration of the incubation media. First, we show that PS exposure was correlated with an increase in [Ca 2ϩ ] i , itself resulting from a stimulation of Ca 2ϩ influx, mediated by ATPgated channel P2X7. In our experimental conditions, we confirm that ATP ec did not induce any Ca 2ϩ release (Refs. 3, 24, 35, and 37 and the present data). Consequently the store-operated Ca 2ϩ influx, previously implicated in PS exposure in erythroleukemia cells treated with Ca 2ϩ ionophore (34), could not be involved in ATP ec -induced PS exposure. Second, in a medium devoid of Ca 2ϩ , PS externalization was correlated with the external Na ϩ concentration. This effect is attributable to Na ϩ influx mediated by ATP-gated channel P2X7 and expected to increase intracellular Na ϩ concentration. The role of Na ϩ influx in PS exposure was confirmed by using the Na ϩ ionophores gramicidin D or monensin. Third, at a 2 mM external Ca 2ϩ concentration, the physiological Ca 2ϩ level in biological fluids, and in the absence of Na ϩ , ATP ec induced PS exposure only in a fraction of the ATP ec -responsive cells as determined in the complete medium containing the physiological Na ϩ concentration. Altogether these data show that ATP ec was able to induce PS exposure by a process dependent solely on Na ؉ influx and that Ca 2ϩ , at physiological concentrations, was not sufficient, in the absence of Na ϩ , to elicit the full PS response in the entire responding cell population.
The mechanisms by which intracellular Ca 2ϩ and Na ϩ induced PS exposure and phospholipid scrambling remain hypothetical. As already described for Ca 2ϩ (6,7), both cations could act on the same target by directly activating a phospholipid transporter, such as scramblase, and inhibiting APLT. Alternatively both cations could modulate some regulators of phospholipid transport. For example, a depletion in intracellular ATP known to inhibit the activity of APLT (10, 50) has been observed and implicated in PS exposure during apoptosis (18,46,51). ATP depletion could result from its consumption by Na ϩ /K ϩ or Ca 2ϩ pumps as a result of their activation to remove Na ϩ and Ca 2ϩ from the cells (46). The hypothesis of APLT inhibition following ATP ec treatment could not be directly ver-ified by measuring NBD-PS uptake since this movement will include both rapid PS scrambling and APLT activity. However, even if it was not inhibited, APLT activity was insufficient to overcome the extremely fast PS randomization induced by AT-P ec . When, after a first application, ATP ec was chelated with Mg 2ϩ , NBD-PC scrambling and PS exposure persisted when they were induced by high Ca 2ϩ concentrations. Cytosolic Ca 2ϩ remained sufficiently elevated to maintain scrambling and APLT inhibition at least during the period of observation. On the contrary, in Na ϩ medium, ATP ec chelation stopped scrambling, allowing relocation of endogenous PS. Following ATP ec chelation, intracellular Na ϩ could be actively extruded through the Na ϩ /K ϩ pump, returning to a concentration below the threshold level required to maintain scrambling. Thus, APLT could be directly inhibited either by Na ϩ or by ATP consumption by the Na ϩ /K ϩ pump. Interestingly the presence of external PS in a subpopulation of sickle cells was correlated with APLT inhibition rather than with an elevated [Ca 2ϩ ] i (52). Furthermore this cell population had a low K ϩ and high Na ϩ content (52) in agreement with a potential role of Na ϩ in PS exposure and APLT inhibition. Although the possibility of regulation of PS externalization in apoptosis by modulating APLT activity is attractive (18,46,51,52), it cannot account for the attached scrambling of the other phospholipids. Consequently the most likely hypothesis could be that Na ϩ , as already described for Ca 2ϩ , acts by both inhibiting APLT and activating scramblase or other molecular complexes able to promote phospholipid transport. A role for intracellular Na ϩ has been proposed to explain other cellular processes such as reverse transport of dopamine (53), stress-activated protein kinase/c-Jun N-terminal kinase activation (54), P2X7-mediated membrane blebbing (36), and platelet procoagulant properties (49). In apoptosis, an early Na ϩ influx is involved in the control of cell shrinkage and activation of the apoptotic program (55). However, the mechanisms by which Na ϩ exerts its action remain largely unknown and would deserve to be investigated.
Here we present evidence that ATP ec induces early PS externalization and phospholipid scrambling, which precede other events of apoptosis, in a maximum of 35-40% of thymocytes. Phospholipid movements are dependent on an increase in intracellular Ca 2ϩ and/or Na ϩ concentrations resulting from activated P2X7 channel. We suggest that Na ϩ , rather than Ca 2ϩ , is the major mediator of the early PS externalization induced by ATP ec . The role of Na ϩ in PS externalization, and consequently cell elimination, would require to be investigated in other physiological or pathological conditions, particularly during apoptosis in which the role of Ca 2ϩ in mediating delayed PS exposure has been challenged.