Regulation of Phosphatidylserine Transbilayer Redistribution by Store-operated Ca2+ Entry

The phosphatidylserine transmembrane redistribution at the cell surface is one of the early characteristics of cells undergoing apoptosis and also occurs in cells fulfilling a more specialized function, such as the phosphatidylserine-dependent procoagulant response of platelets after appropriate activation. Although an increase in cytoplasmic Ca2+ is essential to trigger the remodeling of the plasma membrane, little is known about intracellular signals leading to phosphatidylserine externalization. Here, the role of store-operated Ca2+ entry on phosphatidylserine exposure was investigated in human erythroleukemia HEL cells, a pluripotent lineage with megakaryoblastic properties. Ca2+ entry inhibitors (SKF-96365, LaCl3, and miconazole) inhibited store-operated Ca2+ entry in A23187- or thapsigargin-stimulated cells and reduced the degree of phosphatidylserine externalization concomitantly, providing evidence for a close link between the two processes. In cells pretreated with cytochalasin D, an agent that disrupts the microfilament network of the cytoskeleton, store-operated Ca2+ entry and phosphatidylserine externalization at the cell surface were inhibited. In a context where most of the key actors remain to be identified, these results provide evidence for the implication of both store-operated Ca2+ entry and cytoskeleton architectural organization in the regulation of phosphatidylserine transbilayer migration.

mediated cytosolic Ca 2ϩ signals involves two interdependent and closely coupled events: (i) a rapid and transient release of Ca 2ϩ stored in the endoplasmic reticulum, (ii) followed by slowly developing extracellular Ca 2ϩ entry (1,3,4).
In many cell types, including human platelets, depletion of the intracellular Ca 2ϩ stores induces entry of Ca 2ϩ across the plasma membrane (5), referred to as capacitative or storeoperated Ca 2ϩ entry (SOCE) (6). Cell stimulation by Ca 2ϩmobilizing agents triggers the migration of phosphatidylserine (PS) to the exoplasmic leaflet (7), an aminophospholipid sequestered in the inner leaflet of the plasma membrane of nonstimulated cells (8), followed by the shedding of small vesicles released from the plasma membrane (9,10). These microparticles contain surface proteins and cytoplasmic components of the original cell. Thus, shedding of PS-expressing microvesicles appears to be closely associated with cell surface externalization of PS (11). The exposure of PS is a key step for the hemostatic response (11) and becomes a recognition signal for the phagocytosis of apoptotic cells (12). It has been demonstrated that exposure of PS in erythrocytes and platelets (13) as well as during apoptosis (14) is dependent on extracellular Ca 2ϩ and that a continuously elevated [Ca 2ϩ ] i level may be necessary to sustain the externalization process (15). Furthermore, Ca 2ϩ influx across the plasma membrane and maximal exposure of aminophospholipids seem necessary for membrane microparticle formation in platelets (10). In a previous study (16), we have shown that an alteration of SOCE is correlated with a defective exposure of PS in cells isolated from a patient with an extremely rare inherited disorder of PS externalization, the Scott syndrome. Nevertheless, little information regarding the upstream and downstream intracellular signals leading to PS externalization or phospholipid scrambling has been provided. Although several hypotheses have been proposed regarding direct or indirect coupling mechanisms of the endoplasmic reticulum and the plasma membrane (17,18), little is known about the intracellular signals governing SOCE or the Ca 2ϩ channels mediating this particular form of Ca 2ϩ entry. Recently, Rosado et al. (19) suggested that SOCE is mediated by actin cytoskeleton in human platelets.
To examine the significance of SOCE on PS externalization, we have combined the detection of PS exposure and the measurement of [Ca 2ϩ ] i in human erythroleukemia (HEL) cells, a megakaryoblastic cell line expressing specific receptors found in platelets (20,21). We report here that SOCE regulates PS externalization, and cytoskeletal elements modulate both SOCE and PS transmembrane migration. The present study opens a new field of investigation of possible relationships between Ca 2ϩ signaling and the remodeling of the plasma membrane.
* This work was supported in part by grants from the Institut National de la Santé et de la Recherche Médicale and the Université Louis Pasteur. 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.
Assessment of PS Exposure by Prothrombinase Functional Assay-Cell suspensions were centrifuged at 600 ϫ g for 5 min, washed once in Hanks' balanced salt solution, and resuspended at the concentration of 1 ϫ 10 5 cells/ml. Cells were preincubated at 37°C for 15 min with 3 and 10 M SKF-96365 or 10 M miconazole and for 30 min with 10 M CytD. Cells were then stimulated with 2 M ionophore A23187 for 10 min at 37°C, and centrifuged at 12,000 ϫ g for 2 min to separate the microparticle-containing supernatant from the corresponding cells. Procoagulant PS exposure in stimulated cells and derived microparticles was detected, in the presence of 1 mM CaCl 2 , using a human prothrombinase assay in which this phospholipid is the rate-limiting parameter promoting the activation of prothrombin by factor Xa in the presence of factor Va (27). Thrombin generated by the assembled prothrombinase complex was measured using a chromogenic assay, as described elsewhere (23).
Measurements of [Ca 2ϩ ] i by Flow Cytometry-HEL cells were used at 10 6 cells/ml. They were loaded with 3 M Fluo-3 for 30 min at room temperature and then washed twice in RPMI 1640 medium supplemented with CaCl 2 (final Ca 2ϩ concentration ϳ 1 mM) and resuspended at the same concentration (10 6 cells/ml, 500 l/analysis). To study Ca 2ϩ release and Ca 2ϩ entry separately, the experiments were performed in the presence of EGTA (1 mM), able to chelate the totality of Ca 2ϩ in RPMI 1640 medium. Cells were then stimulated by the different agents, and CaCl 2 was added to restore an extracellular Ca 2ϩ concentration of ϳ1 mM, allowing capacitative Ca 2ϩ entry. Fluo-3 fluorescence was monitored using a FACScan Becton-Dickinson flow cytometer and the CELLQuest software. A baseline value was obtained for each sample by fluorescence measurement for 30 s before addition of pharmacologic agents. Collection was immediately resumed, then terminated after additional 5 min (ϳ50,000 events). Fluo-3 fluorescence was expressed in arbitrary fluorescence intensity units and plotted as FL-1 versus time. To convert these values into absolute [Ca 2ϩ ] i , calibration was performed at the end of each experiment. [Ca 2ϩ ] i was calculated using the equation: where K d is the dissociation constant of the Ca 2ϩ ⅐Fluo-3 complex (400 nM) (28). F max represents the maximum fluorescence (obtained by treating cells with 10 M A23187), and F min corresponds to the minimum fluorescence (obtained for ionophore-treated cells in the presence of 1 mM EGTA). F is the actual sample fluorescence. Fluorescence intensities were expressed as the increase in fluorescence with respect to baseline fluorescence intensity before stimulation.
Flow Cytometry Analysis-PS probing was achieved using annexin V FITC (140 nM final concentration). Incubation at room temperature was allowed to proceed for 10 min before data acquisition. In some cases, cells were activated in the presence of 1 mM CaCl 2 with 2 M Ca 2ϩ ionophore A23187 for 10 min at 37°C before addition of annexin V FITC . In another set of experiments, cell-derived microparticles were separated from activated cells by centrifugation at 400 ϫ g for 5 min at room temperature. Under this condition, 100% of cells were eliminated as well as the larger cell fragments. Annexin V FITC wad added to the microparticle-containing supernatant. Samples were analyzed using the FACScan flow cytometer. The forward light scatter and fluorescence channels were set at logarithmic gains. The forward light scatter setting was E-01. The identification of cell and microparticle populations was performed as previously already described (29).
Statistical Analysis-Results were expressed as mean Ϯ S.E. of at least four separate experiments performed at different culture stages and/or passages, with no obvious difference attributable to the culture conditions. Paired Student's t test was used for the statistical analysis. A p value of Ͻ0.05 was considered significant.

Changes of [Ca 2ϩ ] i -To investigate the changes of [Ca 2ϩ
] i , HEL cells were loaded with the Ca 2ϩ -binding fluorophore Fluo-3, and emission fluorescence shifts were monitored by flow cytometry. As shown in Fig. 1, stimulation by the Ca 2ϩionophore A23187 evoked Ca 2ϩ signal with two characteristic phases: a rapid rise in [Ca 2ϩ ] i within a few seconds, followed by a sustained phase of elevated [Ca 2ϩ ] i lasting several minutes, which resembles Ca 2ϩ entry. To determine whether this sustained Ca 2ϩ phase reflected Ca 2ϩ entry, cells were treated with SKF-96365, an agent known to block store-operated Ca 2ϩ channels (30) and other Ca 2ϩ channels. SKF-96365 by itself significantly increased (p Ͻ 0.05) basal fluorescence intensity, [Ca 2ϩ ] i basal values being 43 Ϯ 5 nM, 87 Ϯ 10, and 103 Ϯ 27 nM in the absence and in the presence of 3 and 10 M SKF-96365, respectively. At 1 M, SKF-96365 did not significantly modify Ca 2ϩ response induced by calcium ionophore A23187 (Table I) reductions of ϳ13 and 20%, respectively (Table I). Above 10 M SKF-96365, cytotoxicity precluded further observations (not shown).
In platelets, A23187 has been reported to induce Ca 2ϩ release from intracellular stores, and after sustained depletion, voltage-independent Ca 2ϩ influx across the plasma membrane becomes activated (7). To confirm which is the pathway affected by SKF-96365 in HEL cells, the same experiments were repeated in the presence of LaCl 3 , an inhibitor of divalent cationic channels (31), and miconazole, an imidazole compound able to inhibit SOCE (32). No significant difference was detected in the basal fluorescence intensity after LaCl 3 treatment ([Ca 2ϩ ] i basal values were 43 Ϯ 6 and 48 Ϯ 12 nM, in the absence and in the presence of LaCl 3 , respectively). However, miconazole increased significantly (p Ͻ 0.05) the basal fluorescence intensity ([Ca 2ϩ ] i basal value was 76 Ϯ 10 nM). As illustrated in Table I, LaCl 3 (30 M) or miconazole (10 M) treatment reduced Ca 2ϩ entry by ϳ20 and 15%, respectively. Collectively, these data suggest that the sustained phase of elevated [Ca 2ϩ ] i induced by A23187 in HEL cells is due to SOCE activation.
Effect of Ca 2ϩ Entry Inhibitors on PS Externalization-In the absence of stimulation ( Fig. 2A), HEL cells and the corresponding supernatants showed a basal PS-dependent prothrombinase activity of 0.01 Ϯ 0.005 and 0.02 Ϯ 0.008 NIH units of thrombin generated per min/ml/2 ϫ 10 5 cells, respectively. Following treatment with 3 and 10 M SKF-96365, these activities increased ϳ2-fold, testifying to a weak ability of this agent to perturb PS asymmetric distribution by itself. After Ca 2ϩ ionophore treatment, prothrombinase activity was ϳ12fold and ϳ3.5-fold enhanced in cells and corresponding supernatants, respectively. SKF-96365 significantly (p Ͻ 0.05) affected the development of prothrombinase activity induced by Ca 2ϩ ionophore with an inhibition of ϳ40 and 48% in the presence of 3 and 10 M SKF-96365, respectively. The same behavior was observed in the corresponding supernatants.
Prothrombinase activity measurements were also performed in the presence of LaCl 3 and miconazole. LaCl 3 interfered in this assay, probably because it was able to form a coordinate complex with PS (33), making the latter inaccessible for the assembly of the coagulation factors at the cell surface. As shown in Fig. 2B, miconazole also decreased the development of prothrombinase activity induced by A23187 in HEL cells; however, the inhibition (34%, p Ͻ 0.05) was lower than that observed with 10 M SKF-96365. Prothrombinase activity was reduced accordingly in the corresponding supernatants.
To establish whether microparticles from HEL cells released after A23187 treatment bear exposed PS at their surface, cells and/or corresponding supernatants were incubated with annexin V FITC , a widely used probe for PS (Fig. 3). Unstimulated HEL cells presented a basal level of fluorescence concerning ϳ75% of the cells (Fig. 3B, upper left region), testifying to the probable weak binding of annexin V FITC to various membrane components, including phospholipids other than PS, because interaction may be favored in the presence of Ca 2ϩ at concentration as high as 1 mM (34). The remaining ϳ25% of events in the upper right region of Fig. 3B likely represents spontaneously activated and/or apoptotic cells expressing a proportion of PS at their surface, with the corresponding PS-positive (i.e. annexin V FITC -positive) microparticle population in the lower right region of Fig. 3E. In agreement with prothrombinase assay (Figs. 2A, 2B, and 4C), A23187 induced extensive PS exposure in cells (Fig. 3C), as shown by the marked shift of ϳ99% of the cells to the upper right region, i.e. that of the highest fluorescence intensities. This was indeed also the case for microparticle shedding (Fig. 3F), as evidenced by the large proportion of annexin V FITC -positive events in the lower right region (91 Ϯ 2%). Hence, microparticles virtually stemming from activated cells bear accessible PS and are therefore procoagulant.

TG-evoked Ca 2ϩ
Influx-To establish whether the Ca 2ϩ entry inhibitors affect Ca 2ϩ release from intracellular stores or SOCE, HEL cells were initially stimulated with TG (1 M), an inhibitor of reticulum endoplasmic Ca 2ϩ -ATPases, in the presence of EGTA (1 mM) in the extracellular medium (Fig. 4A). Under these conditions, the addition of 1 M TG to Fluo-3loaded cells evoked a transient and slow elevation in [Ca 2ϩ ] i due to the release of Ca 2ϩ from internal stores. Subsequent reintroduction of 1 mM CaCl 2 in the external medium induced a sustained increase in [Ca 2ϩ ] i , indicative of SOCE. At 1 M, SKF-96365 had no effect on CaCl 2 -evoked response in TGexposed cells (not shown). Relative to controls, the magnitude of SOCE upon CaCl 2 addition was significantly reduced (ϳ20%) in 3 or 10 M SKF-96365-treated cells after stimulation with 1 M TG (Fig. 4, A and B).
As shown above with SKF-96365, both LaCl 3 and miconazole did not affect Ca 2ϩ release from intracellular stores (Table II). After CaCl 2 reintroduction, these inhibitors were able to significantly reduce SOCE (ϳ19 and 12%, respectively).
Disruption of the Actin Cytoskeleton Inhibits SOCE and PS Exposure in HEL Cells-HEL cells were preincubated with 10 M CytD, a widely used membrane-permeant inhibitor of actin polymerization, which binds to the barbed end of actin filaments (35). Under these conditions, basal values of [Ca 2ϩ ] i were 50 Ϯ 5 and 133 Ϯ 34 nM, in the absence and in the presence of CytD, respectively, but this difference was not statistically significant. CytD reduced A23187-induced Ca 2ϩ signal by ϳ20% (Fig. 5A). As illustrated in Fig. 5B, in the presence of EGTA, CytD reduced TG-evoked Ca 2ϩ signal (by 32%, p Ͻ 0.05), indicating that Ca 2ϩ stores are affected by the inhibition of actin polymerization. In addition, CytD reduced CaCl 2 -evoked response in TG-exposed cells (by 21%, p Ͻ 0.01). Fig. 5C shows the effect of CytD on PS externalization at the HEL cell membrane surface. CytD by itself did not significantly modify prothrombinase activity. However, CytD treatment reduced A23187-induced PS exposure by 38% (p Ͻ 0.05), suggesting that PS transmembrane movements are, at least in part, modulated by actin polymerization. The decrease of prothrombinase activity in the corresponding supernatants was weaker.

DISCUSSION
The present study provides evidence that PS externalization is related to Ca 2ϩ entry in HEL cells, and more specifically to a particular way of Ca 2ϩ entry, referred to as SOCE. In addition, we show that both events, SOCE and PS exposure, are regulated, at least in part, by actin cytoskeleton.
Like platelets and many other nonexcitable cells (5,36), HEL cells undergo SOCE, as observed here after CaCl 2 -induced response in TG-treated cells. Three common inhibitors of SOCE, SKF-96365, LaCl 3 , and miconazole (30 -32, 37), partially inhibited capacitative Ca 2ϩ entry in HEL cells without affecting Ca 2ϩ release from intracellular stores. At the actual concentrations used, miconazole inhibited less efficiently A23187-and TG-induced Ca 2ϩ entry compared with SKF-96365 and LaCl 3 . However, in several other studies (37,38), it has been observed that the inhibitory effect of these compounds differs in magnitude depending on the cell type and concentration.
Little is known about the intracellular signals leading to PS externalization. Our results from the present and previous (16) studies strongly suggest that an alteration of SOCE is correlated with a reduced PS exposure. Indeed, Ca 2ϩ entry inhibitors, SKF-96365 and miconazole, which inhibit store-operated Ca 2ϩ channels, are also able to significantly reduce PS exposure. Ca 2ϩ entry inhibitors also reduced prothrombinase activity in the supernatants, suggesting a Ca 2ϩ dependence for microvesicle release. These inhibitors were deliberately used at low concentration to prevent artificial perturbation of the lipid organization in the plasma membrane, because this may happen in the presence of membrane-binding molecules (8).
After stimulation, HEL cells and microparticles were both strongly labeled by annexin V FITC , confirming that a major fraction of microparticles bore exposed PS (91 Ϯ 2%). Similar results were obtained in platelets (10,39,40), red blood cells (41), monocytes (29), and endothelial cells (42). This suggests that most if not all shed vesicles express PS at their surface and may therefore disseminate membrane-associated procoagulant activity.
The actin microfilaments of the cytoskeleton form a complex network, providing the structural basis for simultaneous interactions between multiple cellular structures. For example, the actin-based cytoskeleton is involved in the control of ion chan-nel activity across the plasma membrane of different cell types (43,44). In the present study, CytD treatment inhibited A23187-induced Ca 2ϩ responses and both Ca 2ϩ release and entry evoked by TG. Similar observations have been reported in astrocytes after agonist stimulation, suggesting that disruption of microfilaments inhibits receptor/G i -protein interaction (45). However, here we can exclude this hypothesis, because HEL cells were stimulated by two agents, A23187 and TG, involving distinct signaling pathways. It is likely that the cytoskeleton regulates Ca 2ϩ signaling at different stages of Ca 2ϩ release and SOCE. In addition, the results concerning the effect of CytD on Ca 2ϩ release and/or Ca 2ϩ entry are conflicting. Ribeiro et al. (46) and Patterson et al. (47) have observed that CytD did not block TG-induced Ca 2ϩ entry in NIH 3T3 or smooth muscle cells, respectively. Other authors have observed that CytD did not alter the agonist-induced Ca 2ϩ release in vascular endothelial cells, but Ca 2ϩ entry was reduced (48). Moreover, in human platelets, Ca 2ϩ stores released by TG were not affected by CytD, but SOCE was greatly reduced by cytoskeletal depolymerization (19). These discrepancies are probably related to a wide variability in the cytoskeletal intrinsic properties as a function of the cell type. Thus, the fact that cytoskeleton disruption attenuates A23187-and TG-induced Ca 2ϩ release and entry in HEL cells indicates that cytoskeletal integrity or closely associated processes are essential for a normal coupling between Ca 2ϩ stores and SOCE.
The role of cytoskeletal reorganization in PS exposure and microparticle formation has been emphasized by several groups who studied the relationships between both events in platelets. Inhibition of actin polymerization by CytD affected the shedding of membrane vesicles but had no effect on the redistribution of PS between the two leaflets of the platelet plasma membrane (49). Other authors (50 -53) suggest that  platelet cytoskeletal proteins are involved in the regulation of membrane lipid asymmetry. However, in the present study, we have observed that the diminution of Ca 2ϩ influx across plasma membrane induced by CytD results in a significant reduction of PS exposure with almost no effect on microparticle shedding. Here, the partial inhibition suggests that the cytoskeleton plays a modulatory role rather than a mandatory one in this particular Ca 2ϩ signaling pathway regulating PS exposure at the HEL cell surface. Recently, it has been observed that the actin-cytoskeleton is also able to modulate the uptake of membrane-derived microparticles by cells (54). These findings, when added to the results of the present study, suggest that cytoskeletal elements actively participate in the intercellular traffic of membrane proteins and, more generally, in the transcellular exchange of biological information. HEL cells express platelet-specific membrane glycoproteins (26) and possess a strong ability of PS externalization following stimulation. Hence, such cells represent a useful model to investigate platelet properties. Among the latter, PS-dependent procoagulant response has an essential physiologic significance as demonstrated by the bleeding tendency in subjects with Scott syndrome, a genetic defect of PS transmembrane migration (55,56). This study provides new insights into the control of PS transbilayer movements by both SOCE and cytoskeleton organization, and opens a new field of investigation of possible relationships with Ca 2ϩ -dependent scramblase (57), and ATP-binding cassette (ABC) transporters such as multidrug resistance P-glycoproteins (58), MRP1 (59), or ABC1 (60), suspected to act as lipid translocases.