Epstein-Barr Virus Latent Membrane Protein 1 Increases Calcium Influx through Store-operated Channels in B Lymphoid Cells*

Ca2+ signaling plays an important role in B cell survival and activation and is dependent on Ca2+ trapped in the endoplasmic reticulum (ER) and on extracellular Ca2+. Epstein-Barr virus (EBV) can immortalize B cells and contributes to lymphomagenesis. Previously, we showed that the ER Ca2+ content of Burkitt lymphoma cell lines was increased following infection with immortalization-competent virus expressing the full set of EBV latency genes (B95–8). In contrast, infection with an immortalization-deficient virus (P3HR-1) not expressing LMP-1 is without effect. LMP-1 protein expression was sufficient to increase the ER Ca2+ content and to increase the cytosolic Ca2+ concentration ([Ca2+]cyt). In this follow-up study, we showed that the resting [Ca2+]cyt of P3HR-1-infected cells was decreased, implying that EBV not only modified the ER homeostasis but also affected the cytosolic Ca2+ homeostasis. Furthermore, even if the store-operated calcium entry (SOCE) of these cells was normal, the [Ca2+]cyt increase after thapsigargin + CaCl2 stimulation was blunted. In contrast, the resting [Ca2+]cyt of B95–8 infected cells was not changed, even if their SOCE was increased significantly. When expressed alone, LMP-1 induced an increase of the SOCE amplitude and the expression of the protein allowing this influx, Orai1, showing the effect of EBV on SOCE of B cells are mediated by LMP-1. However, other hitherto unidentified EBV processes, unmasked in P3HR-1 infected cells, counteract this LMP-1-dependent increase of SOCE amplitude to impair a general and potentially toxic increase of [Ca2+]i. Thus, EBV infection modifies the cellular Ca2+ homeostasis by acting on the ER and plasma membrane transporters.

Engagement of the B cell receptor induces a massive Ca 2ϩ ion influx from the extracellular space (1). This influx is mainly due to a store-operated calcium entry (SOCE) 3 (also known as capacitative calcium entry or CCE), although other influx pathways do exist in B cells (2,3). Inositol 1,4,5-trisphosphate is synthesized rapidly after B cell receptor cross-linking and allows Ca 2ϩ ion release from the endoplasmic reticulum (1). STIM1, a transmembrane protein of the ER membrane, senses the intraluminal [Ca 2ϩ ] decrease and multimerizes and translocates toward the plasma membrane (1). The C terminus of STIM1 is then able to interact directly with Orai1 proteins present in the plasma membrane, inducing their tetramerization in the functional store-operated calcium channel, also known as Ca 2ϩ release-activated channels (4,5). This calcium influx, together with the Ca 2ϩ release from the ER, induces a massive increase of the cytosolic Ca 2ϩ concentration ([Ca 2ϩ ] cyt ), leading to the activation of key calcium-dependent enzymes involved in cell activation (1,6).
Epstein-Barr virus (EBV), a human gammaherpesvirus can immortalize primary naïve B lymphocytes, leading to the establishment of permanently growing lymphoblastoid cell lines (LCLs). EBV is involved in the pathogenesis of several malignancies, including Burkitt lymphoma, some NK/T lymphomas, Hodgkin lymphoma, nasopharyngeal carcinomas, or posttransplant lymphoproliferative disease (reviewed in Ref. 7). The viral LMP-1 (latent membrane protein-1) is an oncoprotein that has been implicated in the genesis of many of these EBVassociated tumors. LMP-1 displays functional homology with the activated CD40 receptor (8), leading to activation of NK-B, ERK, JNK, and p38 MAPKs, and Akt-PI3K pathways and to the downstream phenotypic changes observed during transformation of resting B cells into LCLs. LCLs express about nine EBV proteins, including LMP-1, whose expression is under the control of the EBNA2 virus protein (9).
In a previous study, using the prototype transforming EBV strain (B95-8) and a nontransforming strain deleted for the EBNA2 gene (P3HR-1), we showed that infection of EBV-negative Burkitt's lymphoma cell lines with the virus is able to modify the expression of the proteins responsible for Ca 2ϩ ions uptake in the ER (10). Thus, the immortalizing EBV strain B95-8 increased expression of the "high" Ca 2ϩ affinity SERCA2 and decreases "low" Ca 2ϩ affinity SERCA3. As a consequence, the amount of Ca 2ϩ ions in the lumen of the ER is increased. In contrast, the nonimmortalizing EBV strain P3HR-1 was without effect on the SERCA expression profile (10). Importantly, infection with the P3HR-1 strain of EBV not * This work was supported in part by INSERM and the Association pour la Recherche contre le Cancer. 1  only resulted in a lack of EBNA2 expression but also to a consequent lack of LMP-1 expression (11). As a major difference between the two EBV strains is the expression of LMP-1, we used an inducible vector coding for LMP-1 to study the effect of LMP-1 alone in the EBV-negative B lymphoma lines. Such experiments revealed that LMP-1 did not alter SERCA2 expression but did decrease SERCA3 expression and caused an increase of Ca 2ϩ sequestration in the ER lumen (10). Expression of LMP-1 also increased the resting [Ca 2ϩ ] cyt .
In this follow-up study, we considered the consequences of these events on the activity of SOCE. As activation of SOCE is directly dependent on Ca 2ϩ ion content of the ER and on [Ca 2ϩ ] cyt , we investigated the calcium influx of various EBVinfected cells or cells expressing only LMP-1 and studied expression of the key SOCE proteins Orai1 and STIM1. We also further elucidated the effects of EBV on Ca 2ϩ ion movement through the plasma membrane. Thus, either EBV strain B95-8 or EBV protein LMP-1 both increased the Ca 2ϩ influx and Orai1 expression, whereas STIM1 expression remained constant. In contrast, the nonimmortalizing EBV strain P3HR-1 is without effect on Ca 2ϩ influx but promotes Ca 2ϩ efflux. The modifications of Ca 2ϩ homeostasis by EBV may be linked to tumorigenesis and altered lymphopoiesis.

EXPERIMENTAL PROCEDURES
Cells-BL-30, BL-41 (12), and BJAB (13) cells are EBV-negative human B lymphoma cells. All of the cell lines were maintained in RPMI 1640 medium (Lonza, Levallois-Perret, France) supplemented with 10% heat-inactivated fetal calf serum and 2 mM L-glutamine at 37°C in a 5% CO 2 -humidified atmosphere. B95-8 immortalized B cells from Orai1-deficient and healthy patients were a kind gift of Dr. Picard and Professor Fischer (Study Center of Primary Immunodeficiencies, AP-HP, Hôpital Necker, Paris, France). Written informed consent was obtained from the parents of the patients. The experiments were conducted after approval was given by the institutional review boards at Necker-Enfants Malades Hospital (Paris, France). Cell reagents were from Lonza (Verviers, Belgium).
Induction of LMP-1 Expression by Tetracycline Withdrawal-BJAB-tTA-LMP-1 cells were grown in complete RPMI medium supplemented with 2 mg/ml G418 and 0.5 mg/ml hygromycin B (both purchased from Sigma-Aldrich) and 1 g/ml tetracycline (Fluka, Steinheim, Germany) as described previously (14). To induce LMP-1 expression, exponentially growing cells cultured in the presence of 1 g/ml tetracycline were washed as follows; after centrifugation, the cell pellet was resuspended in 10 ml of complete medium containing 10% fetal calf serum without tetracycline, transferred into a new 50-ml tube, containing 35 ml serum-free RPMI 1640 medium and pelleted again. This washing step was repeated three times. Thereafter, cells were resuspended in complete RPMI 1640 culture medium without tetracycline at an initial density of 2 ϫ 10 5 cells/ml. During the last one or two passages preceding induction of LMP-1 expression and during induction by tetracycline withdrawal, selection antibiotics were omitted. Expression of LMP-1 was tested by Western blot (10).

Measurement of Intracellular Calcium Concentration-
The intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) was recorded by a fluorimetric ratio technique (15,16). Cells were pelleted and resuspended at a density of 10 6 cells/ml in PBS supplemented with 1 mg/ml bovine serum albumin and incubated in the dark with 4 M Indo-1-AM for 1 h at room temperature under slow agitation. Cells were then washed and resuspended in calciumfree Hepes-buffered saline solution (135 mM NaCl, 5.9 mM KCl, 1.2 mM MgCl 2 , 11.6 mM Hepes, 11.5 mM glucose adjusted to pH 7.3 with NaOH) prior to measurement. After centrifugation, 5 ϫ 10 5 to 10 6 cells were suspended in 3 ml of Hepes-buffered saline solution in a quartz cuvette and inserted into a spectrofluorophotometer (RF-1501 Shimadzu Corp., Kyoto, Japan) connected to a personal computer. A temperature of 37°C was maintained by circulating water from a bath. Ultraviolet light of 360 nm was used for excitation of Indo-1, and emissions at 405 and 480 nm were recorded. Background and autofluorescence of the cell suspension were subtracted from the recordings. The maximum Indo-1 fluorescence (R max ) was obtained by adding 1 M ionomycin to the cell suspension in the presence of 10 mM CaCl 2 . Minimum fluorescence (R min ) was determined following depletion of external Ca 2ϩ by 5 where K d is the apparent dissociation constant of the Indo-1⅐calcium complex (230 nM), and R is the ratio of fluorescence value at 405 nM over the one at 480 nm (15). In all the experiments, cells in the quartz cuvette, were pretreated with 1 M thapsigargin for 10 min to induce the calcium release from the ER and the opening of store-operated channels (16). Next, divalent ions were added (Ca 2ϩ , Ba 2ϩ , or Mn 2ϩ ) according to the designed experiment. In the majority of the experiments, 1 mM CaCl 2 was added to measure the variation of [Ca 2ϩ ] i in response to calcium influx (16). To calculate the Ca 2ϩ influx rate (⌬[Ca 2ϩ ] i ), we measure the variation of [Ca 2ϩ ] i per second during the first 20 s after CaCl 2 addition; under these conditions, the activities of the plasma membrane Ca 2ϩ ATPases (PMCA) and of the Na ϩ /Ca 2ϩ exchanger are largely reduced (and the activity of SERCA is already inhibited by thapsigarigin (TG)). Thus, the ⌬[Ca 2ϩ ] i is only due to the Ca 2ϩ ion entry (16).
Ba 2ϩ Uptake and Mn 2ϩ -induced Indo-1 Quenching Experiments-To study directly the "Ca 2ϩ " influx through store-operated channels, we performed two kinds of experiments with surrogate divalent ions, Mn 2ϩ quenching of Indo-1 and Ba 2ϩ accumulation. Mn 2ϩ ions bind to Indo-1 and quench the 430-nm emission wavelength fluorescence. The decrease of fluorescence is directly dependent on Mn 2ϩ entry through SOC. For measurements, 100 M MnCl 2 was added instead of CaCl 2 . Ba 2ϩ ions pass through SOC but are not pumped by PMCA. As Ba 2ϩ ions are not pumped to the extracellular medium, they accumulate, and this uptake is dependent directly on the activity of SOC. Ba 2ϩ ions do not exert SOC retroinhibition as Ca 2ϩ ions do (17). In our experiments, after 10 min of TG-pretreatment of the cells, 5 mM BaCl 2 was added instead of 1 mM CaCl 2 . As the Ba 2ϩ ⅐Indo-1 complex is less fluorescent than the Ca 2ϩ ⅐Indo-1 complex (18), we performed a new calibration for Ba 2ϩ ions; the resting fluorescence ratio (same wavelengths as for Ca 2ϩ ions) before Ba 2ϩ application was used as R min . R max was obtained at the end of the experiment after addition of 10 mM BaCl 2 and 1 M ionomycin. K d was assumed to be 2 M (18).
[Ba 2ϩ ] i was calculated with the same formula used for [Ca 2ϩ ] i . Resting [Ca 2ϩ ] i was measured in the same bath to which 1 mM CaCl 2 was added, and the values averaged over 10 min. The given values are representative of at least three independent experiments.
Sample Preparation and Western Blot-This was done essentially as described in Ref. 19. Briefly, cells were harvested by centrifugation, resuspended in 1 ml of ice-cold NaCl (150 mM), transferred to round-bottom 2-ml Eppendorf tubes, centrifuged, and washed again by centrifugation with 1 ml of ice-cold NaCl solution. The cell pellet was then resuspended in ice-cold 5% trichloroacetic acid and kept at 4°C overnight. The formed protein precipitate was centrifuged at 12,000 ϫ g for 10 min at 4°C, and the supernatant was aspirated. The protein pellet was dissolved in lysis buffer (20) at 30 mg TCA-precipitated protein pellet/ml lysis buffer on a horizontal shaking platform and stored thereafter at Ϫ80°C. SDS-PAGE (60 g total cellular protein per well) in 10% gels and transfer onto nitrocellulose membranes was done as described in Ref. 20. The presence of equal amounts of total protein per lane was verified by Ponceau S staining and densitometry, as described previously (20,21). LMP-1 was detected by Western blotting using an anti-LMP monoclonal antibody mixture (clone CS.1-4, code no. M0897; DakoCytomation, Glostrup, Denmark), STIM1 by an anti-STIM antibody (Sigma-Aldrich), Orai1 by an anti-Orai1 (ProSci., Inc., Interchim, Montluçon, France). Luminograms were obtained using the ECL reagent kit of Amersham Biosciences and were quantified with the ScionImage software (Scion Corp.). Acquisition and quantitative analysis of protein expression by Western blotting has been described earlier in detail (19 -21).

EBV Modifies Resting [Ca 2ϩ ] cyt of Infected Cell
Lines-In a previous study (10), we showed that the transforming EBV strain B95-8 increased the content of Ca 2ϩ in the ER lumen, whereas the nontransforming EBV strain P3HR-1 was without effect. As the resting [Ca 2ϩ ] cyt is an equilibrium between an influx of Ca 2ϩ ions to the cytoplasm from the extracellular medium and from the internal stores (and especially the ER) and an efflux from the cytoplasm to the extracellular medium or to the internal stores, we assessed the effect of the EBV B95-8 EBV strain-induced increase of ER Ca 2ϩ concentration on the resting [Ca 2ϩ ] cyt in chronically infected BL-30 and BL-41 cells.
As expected, LMP-1 expression is only detected in the B95-8-infected BL-30 and BL-41 cell lines (Fig. 1, A and C). However, surprisingly the stable infection of either cell line by the EBV B95-8 strain had no effect on their resting [Ca 2ϩ ] cyt (86 Ϯ 6 nM versus 91 Ϯ 2 nM and 105 Ϯ 7 nM versus 105 Ϯ 8 nM for EBV-negative and B95.8-infected BL30 and BL-41 cells respectively, Fig. 1, B and D). In contrast, the resting [Ca 2ϩ ] cyt of the cell lines infected by the EBV P3HR-1 strain was decreased significantly in both lines; in BL-30, there was a 43% reduction to 49 Ϯ 1 nM (p Ͻ 0.01), and in BL-41, there was a 31% reduction to 72 Ϯ 4 nM (p Ͻ 0.05, Fig. 1, B and D).
Therefore, in cells infected by EBV, the resting [Ca 2ϩ ] cyt does not appear to relate to the ER calcium content, as the increase of the ER Ca 2ϩ content in B95-8 infected cells has no consequence on the resting [Ca 2ϩ ] cyt . Furthermore, the results with P3HR-1 infected cells imply that a process is taking place to remove Ca 2ϩ ions from the cytosolic compartment (inducing a resting [Ca 2ϩ ] cyt decrease), which is not a result of an increased activity of SERCAs.
EBV Induces Profound Changes in [Ca 2ϩ ] cyt Variation after TG Treatment-Treatment of BL-30 and BL-41 lines with TG, a noncompetitive inhibitor of SERCAs, induced a Ca 2ϩ release by the ER in both cell lines (10). In the absence of extracellular Ca 2ϩ ions, the [Ca 2ϩ ] cyt variation (⌬[Ca 2ϩ ] cyt ) is only due to the Ca 2ϩ release (Fig. 2, A and B; 30 -600 s), which, in turn, allows the opening of the store-operated calcium channels present in the plasma membrane. When 1 mM CaCl 2 is added to the extracellular solution, entry of Ca 2ϩ ions leads to an increase of [Ca 2ϩ ] cyt . As shown in Fig. 2, A   extracellular Ca 2ϩ (Fig. 2, A and B). Thus, in B95-8 infected cells (blue traces), the increase in [Ca 2ϩ ] cyt is higher and faster than in EBV-free cells (Ca 2ϩ influx rate ϫ2 for BL-30 and ϫ10 for BL-41 cells; Fig. 2, C and D). After reaching a peak of ϳ450 and 900 nM, respectively, in BL-30 and BL-41 cells, the [Ca 2ϩ ] cyt decreased (Fig. 2, A and B) but remained higher than in EBVfree cells, especially in BL-41 cells. The area under the curve, which reflects the amount of Ca 2ϩ uptake, was increased significantly by 26 and 172%, respectively, in BL-30 and BL-41 cells (p Ͻ 0.01, Fig. 2, E and F).
In both cell lines, ⌬[Ca 2ϩ ] cyt was blunted in cells infected with the EBV P3HR-1 strain (red traces, Fig. 2, A and B). Thus, [Ca 2ϩ ] cyt reached a maximum of 81 Ϯ 13 nM and 113 Ϯ 5 nM, respectively, in BL-30 and BL-41 cells. (Ca 2ϩ influx rate significantly decreased by 10ϫ and twice, respectively.) The area under the curve, which reflects the amount of Ca 2ϩ uptake, was decreased significantly by 71% in P3HR1-infected BL-30 cells relative to EBV-free BL30 cells (Fig. 2E) and by 29% P3HR1infected BL-41 cells relative to EBV-free BL41 cells (Fig. 2F); in both cases, these differences were significant (p Ͻ 0.01).

Immortalizing EBV Strain Increases Divalent Ion Influx-As
[Ca 2ϩ ] cyt is a constant equilibrium between Ca 2ϩ influx and efflux, we replaced Ca 2ϩ by Mn 2ϩ or Ba 2ϩ ions. These ions function as Ca 2ϩ surrogates for SOC channels but are poorly, if at all, pumped outside by the cell (18); they therefore allow the direct study of the store-operated ion entry.
After a 10-min pretreatment with 1 M TG, the addition of 5 mM BaCl 2 resulted in a slow uptake of Ba 2ϩ ions (Fig. 3, A and B) with an initial rate of 2.  0.05, ϩ399%, Fig. 3B). In cells chronically infected by the P3HR-1 EBV strain, the initial Ba 2ϩ uptake rate was unchanged relative to EBV-free cells of BL-30 (2.2 Ϯ 0.6 nM/s) Mn 2ϩ ions quench Indo-1 fluorescence; thus, an increase in the rate of Indo-1 quenching reflects an increase of Mn 2ϩ influx. As shown in Fig. 3, C and D, the B95-8 EBV strain displayed an increased quenching rate relative to EBV-free cells in BL-30 (ϩ33%) and BL-41 (ϩ180%). In contrast, infection by the P3HR-1 strain had no effect on the quenching rate (Fig. 3, C  and D).
Similarly to the ⌬[Ca 2ϩ ] cyt in Fig. 2, A and B, infection by the immortalizing B95-8 strain had more effect on Ba 2ϩ and Mn 2ϩ uptake in BL-41 than in BL-30 cells. In contrast, the nonimmortalizing P3HR-1 EBV strain did not modify significantly the divalent ion influx in either cell line. Thus, influx of divalent ions (Ca 2ϩ , Ba 2ϩ , and Mn 2ϩ ) through the SOC channels is highly increased in cells infected with the EBV B95-8 immortalizing strain but is unaffected by infection with the P3HR-1 EBV strain.

LMP-1 Expression Increases BJAB Cell [Ca 2ϩ ] i Variation after TG Treatment-
In a previous study, we showed that one protein of EBV, LMP-1, is able to modify the ER Ca 2ϩ homeostasis (10) and could mimic the effects of the B95-8 EBV strain. In contrast, the P3HR-1 EBV strain, not expressing LMP-1 (11), has no effect on ER Ca 2ϩ homeostasis (10). In light of the differences we observed between the two EBV strains on divalent ion influx and with the knowledge that a major difference between these two strains is the expression of LMP-1, we then investigated the direct effect of LMP-1 on divalent ion influx.
The BJAB-tTA-LMP-1 cells carry an LMP-1 transgene under the control of a tetracycline-repressible promoter. When the cells are grown in the presence of tetracycline, LMP-1 expression is repressed, whereas tetracycline withdrawal results in LMP-1 expression at levels similar to that observed in EBVinfected cells (10,14). Fig. 4A shows a Western blot for LMP1 protein expression at 0, 2, and 5 days following removal of tetracycline from cultures of the BJAB-tTA-LMP1 cells.
In presence of tetracycline, when LMP-1 is not expressed (Fig. 4A), the addition of external Ca 2ϩ ions to BJAB-tTA-LMP-1 cells caused a biphasic increase of [Ca 2ϩ ] cyt; comprising of a rapid increase from ϳ60 to 600 -700 nM during the first minute (influx rate of 12 Ϯ 1 nM/s; Fig. 4, B and C), followed by a decay of ϳ40% to an elevated plateau of ϳ450 -500 nM in 3 min. Similar results were obtained on wild type BJAB cells and on BJAB cells transfected with an empty tTA vector (data not shown), indicating that the tetracycline repressor itself has no effect on Ca 2ϩ homeostasis.
Induction of LMP-1 expression by removal of tetracycline from the culture medium for 2 days (Fig. 4A) led to an increase of ⌬[Ca 2ϩ ] cyt After 24 h of induction, no significant difference in ⌬[Ca 2ϩ ] cyt was noticeable, although LMP-1 expression was clearly induced at this time (data not shown). After 48 h of induction of LMP-1, the addition of external Ca 2ϩ caused a clear difference in ⌬[Ca 2ϩ ] cyt relative to uninduced cells; there was a faster and higher increase from ϳ90 to 750 -800 nM within 30 s, followed by only a slight decay to a plateau at ϳ750 nM (Fig. 4B). The ⌬[Ca 2ϩ ] cyt at 3 and 4 days post-induction of LMP-1 (data not shown) were similar to results obtained after 5 days (Fig. 4, B and C). Calculation of the Ca 2ϩ influx rate in the first seconds after adding CaCl 2 was faster after 2 days (23 Ϯ 1 nM/s; ϩ98%) and 5 days (46 Ϯ 2 nM/s; ϩ284%) relative to uninduced cells (Fig. 4C). After 5 days, there was no significant decay of [Ca 2ϩ ] cyt (Fig. 4B). When expressed as area under the curve, the total amount of intracellular Ca 2ϩ uptake was increased by 42 and 50%, respectively, after 2 and 5 days of induction (Fig.  4D).
Thus, the expression of LMP-1 alone allows a significant increase of the Ca 2ϩ influx in B cells and could explain the increase of Ca 2ϩ influx in B95-8-infected BL-30 and BL-41 cells. However, we next performed experiments with surrogate ions to validate the influx increase.
Thus, when LMP-1 is expressed in BJAB cells, the divalent ion influx is faster. These results confirm the difference we observed between Burkitt lymphoma cells infected by the B95-8 or the P3HR1 EBV strains.
LMP-1 Expression Blunts Inhibition of SOCE by Gd 3ϩ and SKF96365-Classical inhibitors of SOCE are lanthanides (such as Gd 3ϩ and La 3ϩ ions) in submicromolar concentrations and SKF96365 (22,23). We next examined the effect of LMP1 on the inhibition of SOCE by Gd 3ϩ or SKF96365. The results in Fig. 6 show the effects on BJAB-tTA-LMP-1 cells, induced or not to express LMP-1, that were pretreated for 10 min with 1 M TG in a Ca 2ϩ -free medium (open bar) and various concentrations of inhibitors that were added 30 s prior to the addition of 1 mM CaCl 2 (black bar).
As shown in Fig. 6A, increasing SKF96365 concentrations inhibit the ⌬[Ca 2ϩ ] cyt in a dose-dependent manner following addition of extracellular CaCl 2 in LMP-1 nonexpressing BJAB cells; 10 M SKF96365 almost totally inhibits the Ca 2ϩ influx. The dose-response curve (Fig. 6E) allows the calculation of an inhibition constant K i of 3 M. Similarly, in cells induced to express LMP-1 for 5 days, the Ca 2ϩ influx is also inhibited by increasing SKF96365 concentrations (Fig. 6B) but with a weaker K i of 33 M (Fig. 6E). In the same way, expression of LMP-1 decreases the ability of Gd 3ϩ ions to block the Ca 2ϩ influx by 10ϫ (K i 160 nM versus 17 nM, Fig. 6,  B, D, and F). Thus, LMP-1 expression confers a decreased sensitivity to two commonly used inhibitors of SOCE.
LMP-1 Expression Induces an Increase of Orai1 Expression-Recently, it was shown that the expression levels of the two main proteins implicated in the formation of the Ca 2ϩ influx in lymphocytes, Orai1 and STIM1, may account for pharmacology and selec- tivity differences of the influx (24). We therefore analyzed the effect of LMP-1 expression on Orai1 and STIM1 expression. As shown for duplicate samples in Fig. 7A, the use of an anti-STIM1 antibody allowed the detection of a single band with the expected molecular mass of ϳ80 kDa by Western blot, and the induction of LMP-1 had no effect on the level of STIM1 expression. [Ba 2ϩ ] cyt measurement and Indo-1 quenching by Mn 2ϩ were done as described in the legend to Fig. 3. Briefly, control BJAB cells, not expressing LMP-1 (non-induced) and LMP-1 expressing BJAB cells were pretreated for 10 min with 1 M TG to allow the opening of the store-operated channels. Then according to the experiment, either 5 mM BaCl 2 (A) or 100 M MnCl 2 (B) was added. The anti-Orai1 antibody detected several bands, with a main band corresponding to a molecular mass of ϳ50 kDa. As the unglycosylated form of Orai 1 has a predicted size of 33 kDa, the 50-kDa band likely corresponds to glycosylated Orai1. In presence of LMP-1, the expression of this Orai1 protein was increased significantly by 66 Ϯ 13% (p Ͻ 0.05, duplicate shown in Fig. 7A) without any effects on other, nonspecific bands (Fig.  7, A and B). Thus, the presence of LMP-1 induces an increase of the expression of the pore channel Orai1, whereas the expression of the protein inducing the opening of the channels is unchanged.
Orai1 is the main channel responsible for the SOCE in B cells (25). It has been shown in T cells that Orai2 and Orai3 could be expressed but play no role in the SOCE (26). However, to confirm that the B95-8 EBV strain and especially LMP-1 do not increase the activity of Orai2 and/or Orai3 channels, we performed experiments on B95-8 immortalized LCLs from a patient bearing the heterozygous double missense A103E and L194P mutations (27). These cells did not express any Orai1 protein, and the SOCE was undetectable. We found that the Mn 2ϩ quenching of Indo-1 was almost abolished in Orai1-deficient cells in contrast to LCLs from healthy volunteers (Fig.  7C). Thus, when cells were immortalized by the B95-8 strain and expressed LMP-1, the presence of Orai1 proteins was absolutely required to observe an increase of SOCE.

DISCUSSION
The major finding of this study is that the EBV and especially its transforming protein LMP-1 are able to drastically modify the Ca 2ϩ influx of stimulated B cell lines. As a complement to our previous work, where we showed that LMP-1 modifies the Ca 2ϩ content of the ER, we now provide mechanistic data on how EBV acts on global cell Ca 2ϩ homeostasis.
Calcium signaling plays a critical role in lymphocyte survival and activation. In the last five years, it has been discovered that two proteins play the main roles in T and B cell Ca 2ϩ entry: Orai1 as the pore of the channel and STIM1 as the opener (1). When these proteins are not functional or are absent due to mutations, the Ca 2ϩ entry (known as a store-operated calcium entry) is impaired, and T and B cells cannot be activated (1). Some viruses, like HIV-1, have targeted the SOCE to disturb the immune response (16).
It is likely that the EBV life cycle is regulated by cellular calcium signaling; ionomycin and A23187, two calcium ionophores that allow the entry of external Ca 2ϩ ions, are inducers of EBV reactivation (28). Notwithstanding the role of Ca 2ϩ in the EBV life cycle, data on whether EBV affects SOCE are incomplete, especially in the light of the recent discoveries regarding Orai1 and STIM1.
In a previous work, we showed that the immortalizing strain of EBV, B95-8, increases the Ca 2ϩ ER amount in contrast to the nonimmortalizing EBV strain P3HR-1 (10). Using a B cell line with inducible expression of LMP-1, we could attribute this difference between the EBV strains to the expression of LMP-1. As the activation of the SOCE is directly linked to the Ca 2ϩ ER content, we sought in the present study to examine the Ca 2ϩ entry in EBV chronically infected cells. We now show that EBV also is able to directly modify the fluxes of Ca 2ϩ ions through the plasma membrane. Conditional LMP-1 expression in BJAB-tTA-LMP-1 cells increased the expression of the Orai1 proteins, which form the pore of the store-operated channels of the plasma membrane, allowing a more pronounced entry of Ca 2ϩ ions. In contrast, expression of the STIM1 protein, which induces the opening of Orai1 channels, is not modified, implying that the effects of LMP-1 on the Ca 2ϩ entry specifically target Orai1. Thus, when LMP-1 is expressed in an EBV-negative B cell line, it drastically modifies the Ca 2ϩ homeostasis of the cell through increasing Ca 2ϩ release from the ER (10) and increasing Ca 2ϩ influx from the extracellular milieu (present study). As a consequence, the resting [Ca 2ϩ ] cyt of LMP-1-expressing cells rises. As the ratio of Orai1 and STIM1 expression is known to modify the kinetics and pharmacology properties of the HEK293 cell SOCE (24), the decrease of sensitivity to Gd 3ϩ and SKF96365 inhibition could be attributed to the observed increased Orai1/STIM1 expression ratio.  Fig. 5B, on a wild type Orai1-expressing LCL from a healthy patient (wt Orai1) and on an Orai1-deficient LCL from a SCID patient (Orai1 def).
However, the use of cell lines infected with EBV strains draws a more complex picture and shows that LMP-1 is not the only EBV protein to act on Ca 2ϩ transporters (Fig. 8). Cells infected with the P3HR-1 EBV strain, which do not express LMP-1, have normal Ca 2ϩ influx (Fig. 8B). However, their resting [Ca 2ϩ ] cyt is decreased significantly, indicating that a process to remove Ca 2ϩ ions from the cytosol is active. This cannot be due to SERCA activity as the expressions of the SERCA2 and SERCA3 are not modified by the P3HR-1 infection (10). Furthermore, when cells are treated with TG, an inhibitor of SERCAs, the increase of [Ca 2ϩ ] cyt following addition of extracellular Ca 2ϩ is blunted, confirming that the Ca 2ϩ ion removal from the cytosol is not linked to the SERCA and is probably due to plasma membrane transporters (possibly plasma membrane Ca 2ϩ ATPases and/or Na ϩ /Ca 2ϩ exchanger). Further experiments are needed to confirm that P3HR-1 induces a Ca 2ϩ efflux increase and to identify the transporters involved.
When cells are infected by the B95-8 EBV strain, which induces the expression of LMP-1, the Ca 2ϩ influx is increased, but the resting [Ca 2ϩ ] cyt is unchanged (Fig. 8C). Thus, we can hypothesize that the LMP-1-induced Ca 2ϩ entry increase is counteracted to abrogate a rise in the resting [Ca 2ϩ ] cyt . Although not formally demonstrated, we hypothesize that the efflux increase observed in P3HR-1 infected cells is also present in B95-8 infected cells and could avoid the increase of resting [Ca 2ϩ ] cyt . Thus, an EBV protein other than LMP-1 is able to increase the activity of the efflux transporters and should be explored. One possible candidate for this effect is LMP-2A, another EBV signaling protein expressed during type III latency, which is known to decrease Ca 2ϩ mobilization of LCLs stimulated by the B cell receptor or CD19 receptor (29).
LMP-1 is known to activate several signaling pathways (reviewed in Ref. 30). The balance of signaling events induced by LMP-1 appears to be dependent upon its level of expression, and this is reflected in downstream biological functions because at high concentrations, LMP-1 inhibits cell proliferation, whereas at lower concentrations, LMP-1 seems to favor cell proliferation. The level of LMP1 influences autophagy (31), which is a well known Ca 2ϩ -dependent process: according to the cell type, autophagy is dependent on Ca 2ϩ release from the ER or dependent on the Ca 2ϩ influx (32)(33)(34). Thus, one consequence of EBV increasing the Ca 2ϩ influx through Orai1 channels could be to favor the autophagy to regulate its proliferative/ cytostatic balance and to modulate immune recognition. As the SOCE is tightly regulated by the cells through many regulatory proteins, it is possible that EBV-encoded proteins in addition to LMP-1 could also be implicated.
Orai proteins have a clear link with cell proliferation in various tissues, and Orai1 is known to play a key role in breast tumor cell migration and metastasis (35). Thus, the difference of Orai protein expression and SOCE pharmacology could constitute a new interesting target for the modulation of virusinduced activation, immortalization of B cells, and associated disorders.