Large Store-operated Calcium Selective Currents Due to Co-expression of Orai1 or Orai2 with the Intracellular Calcium Sensor, Stim1*

The molecular nature of store-operated Ca2+-selective channels has remained an enigma, due largely to the continued inability to convincingly demonstrate Ca2+-selective store-operated currents resulting from exogenous expression of known genes. Recent findings have implicated two proteins, Stim1 and Orai1, as having essential roles in store-operated Ca2+ entry across the plasma membrane. However, transient overexpression of these proteins on their own results in little or no increase in store-operated entry. Here we demonstrate dramatic synergism between these two mediators; co-transfection of HEK293 cells with Stim1 and Orai1 results in an approximate 20-fold increase in store-operated Ca2+ entry and Ca2+-selective current. This demonstrates that these two proteins are limiting for both the signaling and permeation mechanisms for Ca2+-selective store-operated Ca2+ entry. There are three mammalian homologs of Orai1, and in expression experiments they all produced or augmented store-operated Ca2+ entry with efficacies in the order Orai1 > Orai2 > Orai3. Stim1 apparently initiates the signaling process by acting as a Ca2+ sensor in the endoplasmic reticulum. This results in rearrangement of Stim1 within the cell and migration toward the plasma membrane to regulate in some manner Orai1 located in the plasma membrane. However, we demonstrate that Stim1 does not incorporate in the surface membrane, and thus likely regulates or interacts with Orai1 at sites of close apposition between the plasma membrane and an intracellular Stim1-containing organelle.

Store-operated Ca 2ϩ (SOC) 3 influx is the major mechanism for Ca 2ϩ entry in many non-excitable cell types. Despite more than two decades of research, little is known about the activa-tion mechanism for the channels responsible for this type of Ca 2ϩ entry. Recently, based primarily on RNAi screens from either Drosophila or mammalian cells, two proteins have been identified as essential components in SOC influx: Stim1 (1,2), and Orai1 (3,4). Stim1 is thought to act as a sensor for Ca 2ϩ in the endoplasmic reticulum, or in that compartment of the endoplasmic reticulum responsible for signaling to store-operated channels. Zhang et al. (5) proposed a mechanism for Stim1-mediated SOC influx by which Stim1, normally an ER membrane resident protein, is transported to and inserted into the plasma membrane upon Ca 2ϩ store depletion. However, others have suggested that Stim1 may re-localize near the plasma membrane without inserting into the membrane upon Ca 2ϩ -store depletion (2). Mammalian cells may also express a homolog of Stim1, Stim2 (1,2,6), although currently its function in SOC entry is uncertain. Orai1 was first identified by Feske et al. (3) through a combined RNAi screen and analysis of gene mutations in patients suffering from severe combined immunodeficiency. The protein appears to be resident in the plasma membrane (3,4), and its molecular function has not yet been defined. However, Vig et al. (4) suggest that Orai1 (which they designated as CRACM1) could function in the plasma membrane either as a component of the calcium release-activated calcium (CRAC) channel, or as a regulator of CRAC channels.
In the current study, we have sought to determine if Stim1 and Orai1 functionally interact by co-expressing them in HEK293 cells. Surprisingly, we find that the combination of Stim1 and Orai1 results in a substantial increase in SOC entry, suggesting that these proteins are limiting for, and may actually comprise both the activation as well as permeation mechanism for SOC influx. This is to our knowledge the first demonstration of an I crac -like current produced by the ectopic expression of known genes. In addition to Orai1, Orai2 expression with Stim1 also enhanced SOC entry in these cells, however to a lesser extent. Although Orai3 alone or with Stim1 showed no elevation in current or Ca 2ϩ entry, it did rescue the knockdown of Orai1 in HEK293 cells. In addition, we have investigated the movements and distribution of Stim1 and conclude that this protein translocates to the vicinity of the plasma membrane, where it presumably interacts with and activates Orai1, but Stim1 does not incorporate into the plasma membrane. * This work was supported by funds from the Intramural Program of NIEHS, National Institutes of Health. 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. 1 Both authors contributed equally to this work. 2 To whom correspondence should be addressed: NIEHS, National Institutes of Health, PO Box 12233, Research Triangle Park, NC 27709. Tel.: 919-541-1420; Fax: 919-541-1898; E-mail: putney@niehs.nih.gov. 3 The abbreviations used are: SOC, store-operated Ca 2ϩ ; EYFP, enhanced yellow fluorescent protein; DMEM, Dulbecco's modified Eagle's medium; siRNA, small inhibitory RNA; IP 3 , inositol 1,4,5-trisphosphate; I crac , calciumrelease-activated calcium current; TRP, transient receptor potential; ER, endoplasmic reticulum; HEK, human embryonic kidney; RNAi, RNA interference; PBS, phosphate-buffered saline; FACS, fluorescent-activated cell sorter; GFP, green fluorescent protein.

MATERIALS AND METHODS
Cell Culture-HEK293 cells, obtained from ATCC, were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum and 2 mM glutamine and maintained in a humidified 95% air, 5% CO 2 incubator at 37°C. In preparation for cDNA or siRNA transfection, cells were transferred to 6-well plates and allowed to grow to ϳ90% confluence. In preparation for Ca 2ϩ measurements, cells were transferred onto 30-mm round glass coverslips (no. 1 thickness) as a 0.5 ml cell suspension (ϳ400,000 cells/ml) and allowed to attach for a period of 12 h. Additional DMEM was then added to the coverslip, and the cells maintained in culture for an additional 12-36 h before use in Ca 2ϩ measurements.
Plasmids-Full-length Stim1, Orai1, Orai2, and Orai3 cDNA plasmids were purchased from Origene in the pCMV6-XL5 (Stim1 and Orai2) and pCMV-XL4 (Orai1 and Orai3) vectors. The following mutations were made in both the native Stim1 plasmid from Origene as well as Stim1 with the yellow-enhanced fluorescent protein fused to the N terminus, obtained from Tobias Meyer, Stanford University. Single amino acid mutations to the putative EF-hand of Stim1 (D76A, D76N, and E87Q) as well as a multi-amino acid mutant (D76N/D78N) were made by site-directed mutagenesis with the QuikChange site-directed mutagenesis kit (Stratagene). The mutations were all sequence-confirmed. The KIR-GFP plasmid was obtained from Deborah Burshtyn, University of Alberta.
siRNA Knockdown-HEK293 cells were plated in a 6-well plate on day 1. On day 2, cells were transfected with siRNA (100 nM) against Stim1 (Dharmacon) or Orai1 (Invitrogen) using Metafectene (Biontex Laboratories GmbH, Martinsried/Planegg, Germany, 7 l per well), and including siGLO (Dharmacon) as a marker. The sequence of the siRNA against Stim1 was: agaaggagcuagaaucucac; for Orai1 it was: cccuucggccugaucuuuaucgucu. After a 6-h incubation period, the medium bathing the cells was replaced with complete DMEM and maintained in culture. On day 3, siRNA treated cells were transfected with cDNA for Stim1 tagged with EYFP, or EYFP alone, as described below. On day 4, cells were transferred to 30-mm glass coverslips in preparation for Ca 2ϩ measurements as described above, which were performed on day 5 or 6.
cDNA Transfection-HEK293 cells were plated in a 6-well plate on day 1. On day 2, cells were transfected using Lipofectamine 2000 (Invitrogen; 2 l per well) with cDNA (0.5 g/well) for EYFP, Orai1, Orai2, Orai3, or EYFP tagged Stim1 (a gift from Dr. Tobias Meyer) and mutant forms of Stim1. The latter constructs have a mutation in the putative EF-Hand region of the Stim1 molecule (described under "Results"). After a 6-h incubation period, the medium bathing the cells was replaced with complete DMEM and maintained in culture. On day 3, cDNA treated cells were transferred to 30-mm glass coverslips in preparation for Ca 2ϩ measurements as described above, which were performed on day 4 or 5. In general, the concentration of plasmid used was 0.5 g/well, except for EYFP (0.1 g/well) Higher concentrations of plasmid (2.0 g/well) were used in some of the experiments with Orai2 and 3, as indicated under "Results." Single Cell Ca 2ϩ Measurements-Fluorescence measurements were made in HEK293 cells loaded with the Ca 2ϩ -sensitive dye, fura-5F, as described previously (2). Briefly, cells plated on 30-mm round coverslips and mounted in a Teflon chamber were incubated in DMEM with 1 M acetoxymethyl ester of fura-5F (fura-5F/AM, Molecular Probes) at 37°C in the dark for 25 min. For [Ca 2ϩ ] i measurements, cells were bathed in HEPES-buffered salt solution (HBSS: NaCl 120; KCl 5.4; Mg 2 SO 4 0.8; HEPES 20; CaCl 2 1.8; and glucose 10 mM, with pH 7.4 adjusted by NaOH) at room temperature. Nominally Ca 2ϩfree solutions were HBSS with no added CaCl 2 . Fluorescence images of the cells were recorded and analyzed with a digital fluorescence imaging system (InCyt Im2, Intracellular Imaging Inc., Cincinnati, OH). Changes in intracellular Ca 2ϩ are represented by and expressed as the ratio of fura-5F fluorescence because of excitation at 340 nm and 380 nm (F340/F380). Before starting the experiment, regions of interests identifying transfected cells expressing the EYFP fluorescence tag were created by observing cells at a 530-nm emission wavelength and illuminated with 477-nm excitation light. Typically, 20 to 30 cells were monitored per experiment. In all cases, ratio values have been corrected for contributions by autofluorescence, which is measured after treating cells with 10 M ionomycin and 20 mM MnCl 2 .
Electrophysiology-Whole cell currents were investigated at room temperature (20 -25°C) in HEK293 cells using the patch clamp technique in the whole cell configuration. The standard extracellular recording solution contained (mM): 140 NaCl, 3 mM KCl, 1.2 MgCl 2 , 0 -10 mM CaCl 2 , 10 glucose, and 10 HEPES (pH to 7.4 with NaOH). Divalent-free solution (DVF) contained the same as above, except Ca 2ϩ and Mg 2ϩ were eliminated and 0.1 mM EGTA was added. Nominally Ca 2ϩ free (NCF) external solution was also the same as above, except no Ca 2ϩ was added. Fire-polished pipettes fabricated from borosilicate glass capillaries (WPI, Sarasota, FL) with 3-5 M⍀ resistance were filled with (in mM): 145 cesium methanesulfonate, 10 BAPTA, 10 HEPES, and 8 MgCl 2 (pH to 7.2 with CsOH). In some experiments, 100 M inositol 1,4,5-trisphosphate (IP 3 ) was directly added to the intracellular pipette solution, and 1 M gadolinium (Gd 3ϩ ) or 30 M 2-aminoethoxydiphenylborane (2APB) was added to the extracellular recording solution. External solution changes were made using a multibarrel perfusion pencil (Automate Scientific) placed adjacent to the cell under investigation. Voltage ramps (Ϫ100 mV to ϩ100 mV) of 250 ms were recorded every 2 s immediately after gaining access to the cell, and the currents were normalized based on cell capacitance (mean ϭ 14.4 Ϯ 1.2 pF). Leak currents were subtracted by taking an initial ramp current before the store-operated currents developed and subtracting this from all subsequent ramp currents. Access resistance was typically between 5 and 10 M⍀. The currents were acquired using pCLAMP-9.2 (Axon Instruments) and analyzed using Clampfit (Axon Instruments) and Origin 6 (Microcal) software.
Confocal Microscopy-For experiments examining the intracellular distribution of Stim1, HEK293 cells, expressing EYFP-Stim1 or EYFP-Stim1 EF hand mutants, were prepared for confocal microscopy in a similar way as described for Ca 2ϩ measurements. Cells plated on 30-mm round coverslips and mounted in a Teflon chamber were placed on the stage of a Zeiss LSM 510 confocal microscope equipped with a 40ϫ water immersion objective (N.A. 1.2). Images of EYFP were obtained with 488-nm excitation light from an argon ion Laser (Lasos T812 M24), and emission fluorescence selected with a 505-550-nm bandpass filter. Confocal images were collected with a pinhole set at 1 Airy unit (ϳ0.9-m image thickness).
For experiments examining surface expression of Stim1, HEK293 cells that had been transfected with cDNA encoding EYFP-Stim1, EYFP-Stim1-D76A, or EYFP-Stim1 in combination with Orai1 were plated onto glass coverslips and allowed to attach overnight. Cells were washed with Ca 2ϩ -free PBS prior to addition of either PBS containing 1.8 mM CaCl 2 or Ca 2ϩ -free PBS with 2 M thapsigargin. Following a 15-min incubation period, cells were fixed by the addition of paraformaldehyde to a final concentration of 1.6%. Cells were incubated 10 min at room temperature and then washed with FACS buffer. Permeabilized cells were washed once for 3 min with FACS buffer supplemented with 0.1% Triton X-100, followed by additional washes with FACS buffer to remove residual detergent. Cells were then stained with anti-EYFP antibody conjugated to Alexa 647 (Molecular Probes, Eugene, OR). Following three wash steps, coverslips were mounted onto slides using ProLong Gold Anti- fade mounting reagent (Molecular Probes). The mounting agent was allowed to set overnight prior to analysis on a Zeiss LSM 510 UV Meta confocal microscope. Background for antibody staining was determined by setting the gain setting just below the point where fluorescence in the 647-nm channel was visible in HEK293 cells transfected with EYFP cDNA. For intact, EYFP-Stim1-transfected samples, the gain was left at this level throughout the analysis. For permeabilized samples, the gain was adjusted downward until the image was just below the saturation point for all pixels.
Total Internal Reflection Fluorescence Microscopy (TIRFM)-TIRFM was carried out using an Olympus (Melville, NY) IX2-RFAEVA-2 illumination system mounted on an Olympus IX71 inverted microscope as previously described (7). Illumination was provided by a 488-nm argon ion laser (Melles Griot, Carlsbad, CA) directed through a fiber optic cable, and emitted flu-orescence passed through a D525/ 50m filter (Chroma) before being captured by a Photometrics Cascade 512F cooled CCD (Roper Scientific, Tucson, AZ). Laser illumination was not toxic to cells, because HEK293 cells loaded with the Ca 2ϩ dye Fluo-3 did not exhibit aberrant Ca 2ϩ release or signs of degeneration when monitored with this system (data not shown).

Co-expression of Stim1 and Orai1
Results in a Synergistic and Robust Increase in SOC Entry- Fig. 1 shows the results of experiments in which HEK293 cells were transiently transfected with EYFP-Stim1, Orai1, or both. Successfully transfected cells were identified by the fluorescence from EYFP-Stim1, or by cotransfected EYFP for control cells or cells transfected only with Orai1. SOC entry was assessed by examining the magnitude of the [Ca 2ϩ ] i signal upon reintroduction of Ca 2ϩ to cells previously treated with the sarcoplasmic-endoplasmic reticulum ATPase inhibitor, thapsigargin, in the absence of extracellular Ca 2ϩ . Graded additions of extracellular Ca 2ϩ resulted in a graded elevation in [Ca 2ϩ ] i . Overexpression of EYFP-Stim1 had little if any effect on Ca 2ϩ entry assessed in this way. Surprisingly, overexpression of Orai1 significantly inhibited SOC entry. More importantly, co-expression of both EYFP-Stim1 with Orai1 resulted in a substantial increase in store-operated Ca 2ϩ entry (Fig. 1). Similar synergism between expression of Stim1 and Orai1 was observed in experiments utilizing Ba 2ϩ as a surrogate for Ca 2ϩ (not shown). The large Ca 2ϩ entry resulting from co-expression of EYFP-Stim1 and Orai1 was blocked by 1 M Gd 3ϩ or 30 M 2APB, which is the expected pharmacological profile of store-operated channels (8). Finally, co-expression of Stim2 with Orai1 resulted in responses resembling those with Orai1 alone, i.e. inhibition of entry (not shown).
We examined whole cell currents in HEK293 cells transfected with these same constructs. In our hands, wild-type HEK293 cells do not reproducibly show detectable store-operated currents, and in other laboratories, these currents have been described as very small and are inconsistently detectable (4, 9, 10). When we broke into HEK293 cells with pipettes con- taining 10 mM BAPTA and 100 M IP 3 , we saw no consistent development of current in control cells ( Fig. 2A) or in cells transfected with either EYFP-Stim1 (not shown) or Orai1 alone. However, in cells co-transfected with EYFP-Stim1 and Orai1, we observed very large inward currents, of the order of 300 -500 pA with 10 mM Ca 2ϩ outside (Fig. 2, A and B). The currents developed rapidly with BAPTA and IP 3 in the pipette ( Fig. 2A), and after a delay when stores were depleted passively with BAPTA alone (Fig. 2B). These currents showed strong inward rectification and reversed at ϩ50 mV, as expected for a calcium-selective channel (Fig. 2C). The current was fully blocked by 1 M Gd 3ϩ (Fig. 2), which at this concentration is believed to block only store-operated channels (11,12). A hallmark of I crac is that rapid removal of extracellular divalent cations causes a transient increase in current due to initial removal of Ca 2ϩ block (13) followed by a process of depotentiation involving loss of Ca 2ϩ from external modulatory sites (14). As shown in Fig. 2D, the large currents observed in Orai1 ϩ Stim1 expressing cells showed similar behavior upon removal of extracellular divalent cations. Finally, the IP 3 -activated current was transiently augmented by 5 M 2APB and was completely blocked by 30 M of this drug, which is a pharmacological hallmark of I crac (15) (Fig. 2E).
Although we did not consistently observe store-operated currents in our untransfected HEK293 cells, based on the value reported by Vig et al. (4) of 0.5 pA/pF, the large I crac -like currents observed with co-expression of Stim1 and Orai1 suggest an increase in current density of at least a factor of 20.
Effects of Expression of Orai2 and Orai3-In addition to Orai1, mammalian cells express two additional homologous genes, Orai2 and Orai3 (3). To determine if these genes also encoded components of an I crac -like entry mechanism, we coexpressed each of these with Stim1 in the same manner as for Orai1. With this protocol, Orai2, like Orai1, inhibits Ca 2ϩ entry on its own, and substantially augments thapsigargin-activated Ca 2ϩ entry when co-expressed with Stim1 (Fig. 3, A and B); however, the increase in [Ca 2ϩ ] i was consistently less than seen in the Orai1 experiments (Orai1 data are included in Fig. 3A for comparison). As for Orai1, this entry is blocked by 1 M Gd 3ϩ or 30 M 2APB (Fig. 3, C and D).
In the experiments in Fig. 3A, cells were transfected with 0.5 g/well of the Orai2 containing plasmid, a similar concentration to that used for Orai1 experiments. In patch-clamp experiments, we found that currents were inconsistently observed with cells transfected in this manner. We thus increased the concentration of Orai2 plasmid to 2.0 g/well, which resulted in larger [Ca 2ϩ ] i increases (this is actually the concentration used for experiments in Fig. 3C) and consistently observed I craclike currents. Fig. 4 shows that overexpression of Orai2 and Stim1 results in currents somewhat smaller than for Orai1, although with similar properties. The currents were transiently increased in divalent-free solutions, and subsequently underwent depotentiation (Fig. 4B). The IV relationships for the Orai2 ϩ Stim1 currents showed strong inward rectification, typical of I crac -like currents (Fig. 4C). No such currents were observed following transfection with Orai2 alone (not shown).
Expression of Orai3 alone failed to suppress SOC entry (not shown) and co-expression of Stim1 and Orai3 failed to produce increased thapsigargin-induced Ca 2ϩ entry, or store-operated currents, even with 2.0 g/well of plasmid (Fig. 5, A  and B). The failure of Orai3 could mean that this particular protein has some function other than regulation or mediation of Ca 2ϩ entry. Alternatively, it could be poorly expressed, or only function in conjunction with other players, for example in complexes with other Orai family members. However, because Orai3 alone did not suppress Ca 2ϩ entry, we were able to obtain evidence that Orai3 can function in store-operated Ca 2ϩ entry in RNAi rescue experiments. As shown in Fig. 5, C and D, knockdown of Orai1 by RNAi results in substantial abrogation of thapsigargin-activated Ca 2ϩ entry, in confirmation of previous reports (3,4,16). When we expressed Orai3 in cells following Orai1 knockdown, thapsigargin-activated Ca 2ϩ entry was essentially restored to the control level (Fig. 5, C and D). We still did not see significant Ca 2ϩ currents in these cells, consistent with the idea that entry was returned only to the control level. The ability of Orai3 to rescue Ca 2ϩ entry following knockdown of Orai1 may indicate that Orai3 is expressed at a more modest (perhaps physiological) level than Orai1, or that it can function together with limited remaining Orai1. Additional experiments will be needed to distinguish among these, or other possibilities. Nonetheless, the clear augmentation of entry by Orai3 in this particular situation demonstrates that this gene product can also play a role in the generation or regulation of store-operated Ca 2ϩ entry.
In summary, we find that in expression experiments, all three Orai genes can generate or augment store-operated Ca 2ϩ entry with efficacies in the order Orai1 Ͼ Orai2 Ͼ Orai3. How this apparent rank order relates to their function under conditions of physiological expression will require additional study.
Stim1 Movements in Response to Ca 2ϩ Store Depletion-We next sought to address the question of how Stim1 might regulate the complex leading to channel activation. Two ideas have been suggested; according to Liou et al. (2), Stim1 redistributes within the cell, approaching the plasma membrane, but does not incorporate into the plasma membrane. A contrasting idea put forth by Zhang et al. (5) suggests that Stim1 actually traffics to, and is incorporated into the plasma membrane where it interacts with key players in the store-operated Ca 2ϩ entry pathway. Also, Spassova et al. (17) presented evidence for a role of plasma membrane Stim1, but did not address the question of whether Stim1 moves into the plasma membrane upon store depletion, or is present there constitutively.
By use of total internal reflected fluorescence (TIRF) microscopy with HEK293 cells transfected with EYFP-Stim1, we confirmed the observation made by Zhang et al. (5) and by Liou et al. (2) that Stim1 does move to the vicinity of the plasma membrane following depletion of Ca 2ϩ stores with thapsigargin (Fig. 6A). We also carried out confocal imaging studies with this construct; consistent with previous findings (2), EYFP-Stim1 reorganizes from an apparent fibrillar or tubular form into discrete punctae in response to depletion of Ca 2ϩ stores with thapsigargin (Fig. 6B).
We prepared several mutant versions of Stim1 targeting acidic residues in the EF-hand domain: D76A (previously reported in Ref.  (D76N, D78N in Fig. 6C), and all produced constitutive Ca 2ϩ entry (not shown, and see D76A and D76N, D78N in Fig. 7B), again with the exception of D76N, which had dominant negative activity (not shown). In single cell Ca 2ϩ imaging experiments we determined that the EYFP-Stim1 construct can fully rescue Ca 2ϩ entry in cells, which have been treated with siRNA directed against Stim1 (Fig. 7A), and that the EYFP-Stim1-D76A and D76N, D78N mutants activate constitutive Ca 2ϩ entry (Fig. 7B).
We next examined both EYFP fluorescence as well as staining of the cells by an antibody (tagged with a fluorophore, Alexa 647) directed against the N-terminal EYFP, which would be extracellular and accessible to the antibody if the protein were inserted into the plasma membrane (18). However, we found that in fixed cells with intact plasma membranes (i.e. no detergent treatment following fixation) antibody binding was not detected before or after depletion of intracellular Ca 2ϩ stores, with either wild-type or the D76A versions of Stim1 (Fig. 7, C and D). This was not caused by failure of the antibody to recognize the epitope as cells which were permeabilized with Triton X-100 prior to Transfected cells were loaded with the fluorescent calcium indicator fura-5F. Before treatment with thapsigargin (2 M), the HBSS bathing the cells was switched from one containing 1 mM Ca 2ϩ to one that was nominally Ca 2ϩ -free. After the thapsigargin-induced [Ca 2ϩ ] i transient had returned to basal levels, the extracellular calcium concentration was elevated to 1 mM. The traces shown contain averaged traces from three similar experiments, and each experiment is an average of data from 20 -30 cells. B, means Ϯ S.E. of the peak increase upon Ca 2ϩ addition for the experiments in A. C, as described under "Materials and Methods," HEK293 cells were transfected with Orai1-siRNA and siGLO. Two days post-transfection, a subset of these cells were subsequently transfected with cDNA for EYFP, Stim1-EYFP or Orai3 plus EYFP. The concentration of Orai3 cDNA used in these experiments was 2 g per well of a 6-well plate. 48 h after cDNA transfection, cells attached to coverslips were then loaded with fura-5F and treated with thapsigargin in the absence and then the presence of 1.8 mM extracellular Ca 2ϩ . The traces shown contain averaged traces from three similar experiments, and each experiment is an average of data from 20 -30 cells. D, means Ϯ S.E. of the peak increase upon Ca 2ϩ addition for all three experiments in C.
We next performed flow cytometry with fixed, intact Jurkat cells which had been transfected with EYFP-Stim1 and stained with anti-EYFP Alexa 647. This technique allows the separation of viable cells from non-viable cells based on forward and side scatter properties. 99.8% of Jurkat cells that had been transfected with the NK cell receptor KIR, which is expressed with GFP fused with its extracellular domain, (19) stained positive with the antibody (Fig. 8A). In comparison, only 2.25% (ϩ 0.23%) of Jurkat cells that were transfected with EYFP-Stim1 stained with the antibody, and those cells were only weakly stained (Fig. 8A). Importantly, in Jurkat cells that were treated with thapsigargin in nominally Ca 2ϩ free buffer for 15 min to deplete Ca 2ϩ stores prior to staining only 2.8% (ϩ 0.5%) of cells fell into the antibody-positive gate (Fig. 8A). Additionally, the mean fluorescence intensity for these cells is not significantly different from cells transfected with cytoplasmic GFP and which fall into the positive gate (p ϭ 0.459 and 0.352, respectively), indicating that these few cells are likely falsely positive. We did observe a somewhat higher percentage of cells that stained with the antibody when transfected with a dominant active mutant version of EYFP-Stim1 (D76A) (2) that results in activation of constitutive Ca 2ϩ entry (2) (Fig. 8B). However, the average number of positive cells was still very low, only 17.05% (Ϯ4.35%) for resting cells and 20.5% (Ϯ1.8%) for TG-treated cells, while virtually all cells transfected with this construct exhibit constitutive Ca 2ϩ entry. Finally, permeabilization of the transfected Jurkat cells reveals a substantial staining of both EYFP-Stim1 and EYFP-Stim1-D76A (Fig. 8, C and D).
We considered the possibility that only a small fraction of EYFP-Stim1 might need to move to the surface, and this might be difficult to detect against the large background of intracellular fluorescently tagged molecules. In addition, it is apparent from the data in Fig. 7A that knockdown of Stim1 by RNAi is incomplete and thus a small but sufficient complement of surface Stim1 might remain in these cells. Thus we co-transfected cells with the combination of EYFP-Stim1 and Orai1 and attempted to detect surface-expressed EYFP-Stim1 as in Fig. 7. We reasoned that if plasma membrane Stim1 is a necessary component or signal for SOC entry, then the huge increases in SOC entry and current density that result from overexpression of EYFP-Stim1 and Orai1 should be accompanied by a correspondingly significant increase in the levels of EYFP-Stim1 detectable in the plasma membrane. However, the results were indistinguishable from those in Fig. 8; again we failed to observe any antibody labeling of the surface of EYFP-Stim1 ϩ Orai1-cotransfected cells (Fig. 9).

DISCUSSION
Since the inception of the concept of capacitative Ca 2ϩ entry (20), research has focused on two basic questions: the signaling mechanism linking the depletion of intracellular Ca 2ϩ stores to plasma membrane Ca 2ϩ channels, and the identity (or identities) of the store-operated channels (21,22). Various hypotheses have been put forth for the signaling mechanism, including a diffusible messenger, or "calcium influx factor" (CIF) (23,24), and a protein-protein interaction mechanism involving coupling of endoplasmic reticulum IP 3 receptors to plasma membrane Ca 2ϩ channels, sometimes referred to as a "conformational coupling" model (25,26). Perhaps the leading candidates for the channels themselves have been members of the TRP superfamily (27,28). However, despite evidence implicating TRPs from knockdown or knock-out experiments, and despite clear demonstrations that TRPs can, under specific experimental conditions, be activated by Ca 2ϩ store depletion, in no instance has a current been generated by expression of TRPs that resembles the archetypical SOC current, I crac . Thus, it is possible that SOC currents that have distinct properties from I crac may involve channels composed of TRP subunits.
We found little or no effect of overexpression of Stim1 on store-operated Ca 2ϩ entry, and, surprisingly, inhibition of entry by overexpressing Orai1 or Orai2. This latter finding was not investigated in detail in the current study; we speculate that it may reflect the same phenomenon seen with overexpression of TRP channels whereby overexpression of one member of a complex of more than two proteins reduces the likelihood of correctly assembling a functional signaling unit (29). In this instance, the other two components could be either two Stim1 molecules, Stim1 and another as yet unidentified component, or other structural components of the signaling complex (30,31).
Remarkably, Stim1 and Orai1 (or Orai2) appeared to interact synergistically such that a large current with I crac -like properties was generated by overexpression of the two proteins together. While this article was in review, reports by Peinelt et al. (32) and Soboloff et al. (33) appeared describing similar, large currents in HEK293 cells, and a report by Zhang et al. (16) described a similar synergism between Stim and Orai in Drosophila. The evidence that the N-terminal EF hand of Stim1 serves as the sensor for endoplasmic reticulum luminal Ca 2ϩ is strong (2,5,17). We cannot at this stage definitively say that Stim1 acts directly on Orai1, or acts through intermediate effectors, for example through gener-ation of a Ca 2ϩ influx factor (24). However, the size of the store-operated Ca 2ϩ -selective current obtained with these two proteins argues that additional stoichiometric components may not be necessary; thus, these two proteins may be capable of reconstituting both the activation and permeation mechanisms of Ca 2ϩ -selective store-operated channels. Although neither Stim1 nor Orai1 show any obvious sequence homology to known ion channels, I crac itself has biophysical properties quite different from other known ion channels (13,22,34). Further studies will be needed to definitively confirm or refute the hypothesis that any or all of the Orai proteins can function as the major pore-forming constituent of Ca 2ϩ -selective store-operated channels such as I crac . If this is not so, however, it would imply that HEK293 cells constitutively express a huge excess of normally nonfunctional CRAC channels. Nonetheless, despite almost a decade of attempts to produce store-operated currents by expression of, for example, TRP genes, to our knowledge the combination of Stim1 and Orai1 represents the first success- . Cells were loaded with fura-5F/AM and their intracellular Ca 2ϩ content was assessed by single cell Ca 2ϩ imaging. Initially, cells were kept in buffer containing 1.8 mM CaCl 2 , cells were then switched to buffer nominally free of Ca 2ϩ , and finally the buffer was exchanged again with buffer containing 1.8 mM CaCl 2 . Shown are average traces from at least 20 -30 cells per trace. C and D, HEK293 cells were transfected with N-terminally tagged EYFP-Stim1 (C ) or EYFP-Stim1-D76A (D). Before fixation and permeabilization with Triton X-100 (bottom panels only), a subset of cells was treated with thapsigargin for 15 min in Ca 2ϩ -free PBS to deplete intracellular Ca 2ϩ stores. Cells were then stained with anti EYFP Alexa 647-conjugated antibody. Cells were then analyzed by laser scanning confocal microscopy to detect EYFP and/or Alexa 647. ful reconstitution of an I crac -like current by ectopic expression of known genes. 4 There are two other potential homologs of Orai1 in the mammalian genome, Orai2 and Orai3 (3). The homology between the sequences of these proteins and Orai1 suggests they may have similar physiological functions. As a first step to understanding their function, we expressed these proteins with and without Stim1, as with Orai1. Orai2 behaved qualitatively similarly to Orai1; when expressed alone entry was inhibited, and when expressed with Stim1, large entry and currents were obtained. In this particular assay, Orai2 dependent currents were smaller than those obtained with Orai1. With Orai3, we saw no increased Ca 2ϩ signals or currents when expressed with Stim1. However, Orai3 was able to rescue Ca 2ϩ entry in cells in which Orai1 was knocked down by RNAi. Thus we conclude that all three Orai homologs are capable of constituting or regulating store-operated channels. This raises the interesting possibility that native SOC channels may involve combinations of Orai proteins. Future work will be necessary to determine their respective roles in specific physiological environments.
Zhang et al. (5) utilized surface biotinylation of Stim1 to determine that it is inserted into the plasma membrane following Ca 2ϩ store depletion. This technique relies on covalent attachment of cell impermeant biotin to proteins located on the cell surface. Following cell lysis, the biotinylated proteins are separated from other cellular proteins by immobilizing them onto streptavidin-coated agarose beads and washing away nonbound proteins. This technique has been widely used to identify proteins that have access to the extracellular space. However, an important caveat for this technique is that dead or dying cells that have compromised plasma membrane integrity will be permeable to the biotin and thus a portion of biotinylated proteins in the extract can be intracellular proteins from these cells. This could be particularly problematic when the proportion of total protein believed to lie on the surface is very small (as appears to be the case here). In addition, it is possible that proteins that are not actually within the plasma membrane, but bind tightly to plasma membrane proteins can be detected with this technique (for example, see Ref. 35). To circumvent these potential problems, we used surface labeling of HEK293 cells, as well as flow cytometric analysis of surface labeling of Jurkat T-cells; the latter technique permits gating on physiologically intact cells only on the basis of light scattering properties. With both assays, we failed to observe significant localization of Stim1 in the plasma membrane, and we did not observe any movement of immunodetectable Stim1 to the surface in response to Ca 2ϩ store depletion. Because we are examining the distribution of a transfected fusion protein, we cannot definitively rule out the possible presence of a small constitutive quantity of native Stim1 in the plasma membrane, as suggested by Spassova et al. (17). However, our data indicate that stimulated transfer of Stim1 from endoplasmic reticulum into the plasma membrane in response to Ca 2ϩ store depletion does not occur, and that the plasma membrane insertion model for Stim1 is incorrect. We have demonstrated that EYFP-Stim1, which can rescue Ca 2ϩ entry in Stim1 siRNA-treated cells, and EYFP-Stim1(D76A), which produces constitutive Ca 2ϩ entry, are undetectable at the cell surface of intact cells by immunofluorescence microscopy and only weakly detectable in a small percentage of cells by flow cytometry. In the case of the D76A mutant of Stim1, Ca 2ϩ entry is activated in the apparent absence of store depletion such that even if native Stim1 were capable of responding differently from the EYFP fusion, there is no stimulus for its activation (i.e. no store depletion). Furthermore, the fact that we observed no EYFP-Stim1 in the plasma membrane of HEK293 cells under conditions whereby the quantity of functional channels would appear to have been dramatically increased calls into question any significant role for plasma membrane Stim1, whether constitutive or otherwise. These data support a model for re-localization of Stim1 within the ER following Ca 2ϩ store depletion, as suggested by Liou et al. (2). This is an important distinction because it rules out the possibility of Stim1 as a subunit of the store-operated channel, and suggests that the C terminus of Stim1 likely interacts with an intracellular target, perhaps Orai1, which either regulates or forms the store-operated Ca 2ϩ channel.