The Third Transmembrane Segment of Orai1 Protein Modulates Ca2+ Release-activated Ca2+ (CRAC) Channel Gating and Permeation Properties*

Orai1, the pore subunit of Ca2+ release-activated Ca2+ channels, has four transmembrane segments (TMs). The first segment, TMI, lines the pore and plays an important role in channel activation and ion permeation. TMIII, on the other hand, does not line the pore but still regulates channel gating and permeation properties. To understand the role of TMIII, we have mutated and characterized several residues in this domain. Mutation of Trp-176 to Cys (W176C) and Gly-183 to Ala (G183A) had dramatic effects. Unlike wild-type channels, which exhibit little outward current and are activated by STIM1, W176C mutant channels exhibited a large outward current at positive potentials and were constitutively active in the absence of STIM1. G183A mutant channels also exhibited substantial outward currents but were active only in the presence of 2-aminoethoxydiphenyl borate (2-APB), irrespective of STIM1. With W176C mutant channels inward, monovalent currents were blocked by Ca2+ with a high affinity similar to the wild type, but the Ca2+-dependent blocking of outward currents differed in the two cases. Although a 50% block of the WT outward current required 250 μm Ca2+, more than 6 mm was necessary to have the same effect on W176C mutant channels. In the presence of extracellular Ca2+, W176C and G183A outward currents developed slowly in a voltage-dependent manner, whereas they developed almost instantaneously in the absence of Ca2+. These changes in permeation and gating properties mimic the changes induced by mutations of Glu-190 in TMIII and Asp-110/Asp-112 in the TMI/TMII loop. On the basis of these data, we propose that TMIII maintains negatively charged residues at or near the selectivity filter in a conformation that facilitates Ca2+ inward currents and prevents outward currents of monovalent cations. In addition, to controlling selectivity, TMIII may also stabilize channel gating in a closed state in the absence of STIM1 in a Trp-176-dependent manner.

Store-operated Ca 2ϩ entry (SOCE) 3 via the Ca 2ϩ releaseactivated-Ca 2ϩ (CRAC) channels requires interaction of the pore-forming Orai1 protein with the endoplasmic reticulum (ER) -membrane bound luminal Ca 2ϩ sensor, STIM1. Upon intracellular Ca 2ϩ store depletion, STIM1 translocates near the plasma membrane (PM)-proximal ER to activate Orai1 (1)(2)(3). This stimulatory effect involves interaction of a short fragment of the STIM1 C terminus comprising approximately 100 residues and termed CRAC-activation domain, STIM Orai1-activating region, Orai1-activating small fragment, or Ccb9, with two regions of Orai1 delimited by residues 70 to 91 in the N terminus and residues 254 to 301 in the C tail (4 -7). It has been suggested that interaction of this STIM1 fragment with the C terminus domain is essential for stabilization of Orai1 in the PM/ER junctional areas, whereas interaction with the N terminus domain is necessary for channel activation (6).
The biophysical properties of CRAC currents have been studied extensively. The key properties of these currents include a high selectivity for Ca 2ϩ over monovalent ions, low unitary conductance of Ͻ 1pS, and regulation of channel activity by Ca 2ϩ ions (2,3). In the presence of extracellular Ca 2ϩ , the current-to-voltage relationship (I/V) of Orai1-mediated CRAC currents is characterized by large inward currents and negligible outward currents. In the absence of divalent ions, in "symmetrical" Na ϩ containing solutions, the I/V approaches linearity and shows outward currents (8). In addition, CRAC currents are characterized by fast and slow inactivation (9). On the basis of the observation that mutation of Glu-106 and Glu-190 residues, which impair ion selectivity, caused loss of fast inactivation, it was suggested that the inactivation gate and the selectivity filter were associated (10). 2-aminoethyldiphenyl borate (2-APB), a widely used blocker of CRAC currents, also affects ion selectivity and activates the E190Q and E106D mutant channels in the absence of STIM1 (11). These observations led to the similar conclusion that the concomitant effect of 2-APB on gating and ion selectivity could involve a single mechanism where gating and selectivity are coupled. Thus, the selectivity filter or a closely associated structure may act as a channel gate. When coexpressed with STIM1, Orai1 currents are first potentiated and then inhibited by 2-APB. In contrast, Orai3 is directly opened in a STIM1-independent manner by extracellular application of 2-APB (Ͼ 30 M) (12)(13)(14). In this case, the I/V exhibits inward currents at hyperpolarized membrane potentials and large outward currents at potentials more positive than ϩ50 mV (11,12,14,15). The mechanisms responsible for the differences in gating and permeability between Orai1 and Orai3 remain largely unknown. Recent studies using chimeras have shown that an Orai1 chimera containing the intracellular loop between TMII and TMIII and the TMIII derived from Orai3 behaves similar to Orai3 and shows large outward currents upon external application of 50 M 2-APB (14).
In a previous study, we systematically mutated residues in the intracellular loop of Orai1 and found all the mutants to behave like the wild type in terms of inward rectification and STIM-dependent activation (16). In this study, we performed mutagenesis of residues in Orai1 TMIII and identified two amino acids, Trp-176 and Gly-183, the mutation of which confer STIM1 independence and affect channel gating and permeation properties. Although the W176C mutant channel was constitutively active, the G183A mutant channel carried currents only in the presence of 2-APB. We found that both mutations induce large outward currents. Although the binding affinity for Ca 2ϩ at hyperpolarized membrane potentials remained similar to the wild type, a dramatic decrease in Ca 2ϩ blocking efficacy of the outward current was observed in these mutant channels. It has been suggested that the high-affinity Ca 2ϩ -binding site, which is responsible for ion selectivity, comprises a single site formed by Glu-106 in TMI (8). On the basis of these data, we propose a model whereby the negatively charged residues Asp-110 and Asp-112 in the TMI-TMII extracellular loop, which are known to evoke a large outward current at depolarized potentials (13), facilitate local Ca 2ϩ accumulation within the outer vestibule of the pore and prevent Ca 2ϩ release from the Glu-106 Ca 2ϩbinding site at depolarized potentials. Such a mechanism would account for the inwardly rectifying I/V of WT CRAC currents. Accordingly, we propose that the W176C and G183A mutations in TMIII allosterically prevent Ca 2ϩ accumulation near Asp-110/D112, resulting in the expulsion of Ca 2ϩ from the Glu-106 site at depolarized potentials, the generation of large outward currents, and the loss of ion selectivity.

EXPERIMENTAL PROCEDURES
Reagents-Thapsigargin and 2-APB were purchased from EMD Biochemicals. Polyclonal rabbit antibody for detection of Orai1 was generated, affinity-purified (Open Biosystems, Huntsville, AL), and used at 1:1000 dilution for immunocytochemistry. Alexa Fluor 568-labeled secondary antibodies were purchased from Invitrogen and used at 1:1000 dilutions for immunocytochemistry.
Plasmids-Full-length cDNA of human Orai1 subcloned into the bicistronic retroviral expression vector pMSCV-CITE-eGFP-PGK-Puro, which allows simultaneous expression of Orai1, GFP, and a puromycin resistance gene, has been described previously (17). Single-point mutants were generated using the QuikChange XL site-directed mutagenesis kit (Stratagene) following the manufacturer's instructions. All the clones were verified by sequencing. For total internal reflection fluorescence (TIRF) analysis, WT, G183A, and W176C mutant Orai1 cDNAs were fused in frame with pEGFP to have a C-terminal GFP tag. The STIM1-mCherry plasmid has been described previously (18).
Cell Lines and Transductions-HEK293 cells were obtained from the ATCC and cultured in DMEM (Mediatech, Hargrave, VA) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT), 10 mM HEPES, 10 mM Glutamine and 1% penicillin/streptomycin (Mediatech, Hargrave, VA). Cells were transfected at 80 -90% confluency using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. For retroviral transductions, phoenix cells stably expressing gagpol and ecotropic env (purchased from the ATCC) were transfected with plasmids encoding Orai proteins to produce an ecotropic, replication-incompetent retrovirus using the calcium phosphate transfection method. Virus-containing supernatant was collected 2 and 3 days after transfection, and immortalized Orai1 Ϫ/Ϫ murine embryonic fibroblasts (MEFs) (16,19) were transduced twice on day 2 and day 3 in the presence of 8 g/ml polybrene. Transduction efficiencies were evaluated visually by GFP expression and Orai1 expression using immunoblotting.
Single-cell Ca 2ϩ Imaging-Orai1 Ϫ/Ϫ MEFs or HEK293T cells were grown directly on UV-sterilized coverslips and loaded with 2 M fura-2/acetoxy methyl ester for 45 min at 22-25°C. For [Ca 2ϩ ] i measurements, cells were mounted in a RC-20 closed-bath flow chamber (Warner Instrument Corp., Hamden, CT) and analyzed on an Olympus IX51 epifluorescence microscope with Slidebook 5.0 (Intelligent Imaging Innovations, Inc.) imaging software. Cells were perfused with Ca 2ϩfree Ringer's solution, and Ca 2ϩ stores were passively depleted with 1 M thapsigargin. Fura-2 emission was detected at 510 nm with excitation at 340 and 380 nm, and the Fura-2 emission ratio (340/380) was acquired at every 5-s interval after subtraction of the background. For each experiment, 35-50 individual cells were analyzed using OriginPro (Originlab) analysis software. [Ca 2ϩ ] i was estimated from the relation [Ca 2ϩ ] i ϭ K*(R-R min )/(R max -R). K, R min (minimum ratio value), and R max (maximum ratio value) were measured independently in control cells.
Immunocytochemistry-Immunocytochemistry for Orai1 staining was performed as described (16). Briefly, transfected HEK293T cells were fixed with 4% paraformaldehyde for 20 min at room temperature, permeabilized with wash buffer containing 0.5% Nonidet P-40, and incubated with anti-Orai1 antibodies (16) and Alexa Fluor-labeled secondary antibodies. Immunofluorescence was analyzed by epifluorescence microscopy using an Olympus IX51 epifluorescence microscope (Olympus, Inc.) and a Plan Apochromat ϫ40 oil differential interference contrast objective with numerical aperture of 1.3.
TIRF Microscopy Analysis-HEK293 cells were transfected with plasmids encoding STIM1-mCherry along with either WT or mutant Orai1-GFP fusion protein encoding cDNAs at a molar ratio of 1:1. TIRFM was performed using an Olympus IX2 illumination system mounted on an Olympus IX51 inverted microscope. Lasers beams from a 488-nm argon ion laser (Melles Griot) and a 594-nm diode laser (Cobolt Instruments) were combined and controlled using an Olympus OMAC TIRF dual port condenser and controller system. The angle of the incident light at the interface between the glass coverslip and the aqueous medium was controlled by independently adjusting the position of each laser beam before passing through a ϫ60 oil immersion objective (NA 1.49). The emission was filtered either at D525/50 nm or 660/50 nm filter (Chroma) and captured by a Hamamatsu ORCA cooled CCD (Roper Scientific) camera. Acquisition and image analysis were performed using Slidebook (Intelligent Imaging Innovations, Inc.) and OriginPro8.5 software.
Measurement of CRAC Currents by Whole Cell Recording-For recording of CRAC currents, HEK293T cells were cotransfected with plasmids encoding Orai1 WT or mutant cDNAs in the presence or absence of a STIM1-encoding plasmid at a molar ratio of 1:1 using Lipofectamine 2000 (Invitrogen). Cells were used for experiments 24 -48 h post-transfection. Patchclamp recordings were performed using an Axopatch 200B amplifier (Molecular Devices, CA) interfaced to a Digidata 1320A (Axon Instruments, CA) for stimulation and data acquisition. Currents were filtered at 1 kHz with a four-pole Bessel filter and sampled at 5 kHz. Recording electrodes were pulled from borosilicate glass capillaries (WPI, Sarasota, FL) using a Flaming Brown pipette puller (Sutter Instrument, CA) to a final resistance of 2-7 M⍀. Stimulation, data acquisition, and analysis were performed using pCLAMP8 and SigmaPlot or Origin software. The standard extracellular Ringer solution contained 145 mM Cs-aspartate, 4.5 mM KCl, 6 mM CaCl 2 , 10 mM D-glucose, and 10 mM Na-Hepes (pH 7.35). The standard internal solution contained 145 mM Cs-glutamate, 8 mM MgCl 2 , 12 mM EGTA, and 10 mM Cs-Hepes (pH 7.3). Experiments described in Figs. 2 and 3 were performed using symmetrical Na ϩ -containing solutions with extracellular Na-aspartate and intracellular Na-glutamate. Unless stated otherwise, the cell membrane was held at 0 mV, and pulses were applied between Ϫ110 mV to ϩ115 mV at 15-mV intervals for 250 ms. We used HEDTA/ Ca 2ϩ for the extracellular solutions to obtain varying amounts of free Ca 2ϩ , depicted in Figs. 3 and 6. For each solution, the concentration of HEDTA and Ca 2ϩ was calculated using the WEBMAXC program and validated using a Ca 2ϩ electrode.
Analysis of Patch Clamp Data-Ionic currents from cells expressing WT and mutant channels were recorded using only the analog compensation of the membrane linear components. In most cases (mutant channels), 2-APB had no inhibitory effects, and the current traces could not be corrected for leak currents. The time course of outward current activation was fitted by a single exponential function. The first 1 ms of recorded data following the onset of the voltage pulses was not included in the fitting to minimize the effect of uncompensated membrane capacitance on the estimated time course of the current. For the I/V, we used steady-state currents measured at the end of the pulse unless specified otherwise.

Analysis of TMIII Mutants of Orai1 for Store-operated Ca 2ϩ
Entry-To determine the role of TMIII in the permeation properties of CRAC channels, we mutated a series of residues in TMIII and examined their properties (Fig. 1, A and B). All the mutants examined in TMIII are highlighted in red in Fig. 1A. Among these mutants, E190Q has been characterized previ-ously and shows altered ion selectivity (15,20). To measure the activity of individual mutants without interference from endogenous Orai1, mutants were retrovirally expressed in Orai1-deficient (Orai1 Ϫ/Ϫ ) MEFs (19). SOCE was strongly reduced in Orai1-null MEFs but was restored to WT levels by expression of WT Orai1 ( Fig. 1B and Ref. 19). The W176C mutant showed very high levels of basal intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ), whereas mutants S179C and G183A did not reconstitute SOCE in Orai1-null MEFs. All the other mutants (E173C, A175C, A177C, F178C, C195G, and W196C) showed normal Ca 2ϩ entry as measured by ratiometric Ca 2ϩ -imaging and current recordings ( Fig. 1B and data not shown). All the mutants showed comparable expression levels as judged by immunoblotting (Fig. 1C) and showed plasma membrane localization similar to wild-type Orai1 ( Fig. 1D and supplemental Fig. 1). These studies suggested that the Ser-179 and Gly-183 residues play an important role in channel gating and that the W176C mutant may be constitutively active and independent from store-depletion.
The W176 Residue Is Important for Gating and Permeation Properties of CRAC Channels-To examine the biophysical profile of currents generated by W176C mutant channels, we expressed W176C mutant channels with and without STIM1 in HEK293 cells. As reported previously (21)(22)(23), coexpression of wild-type Orai1 and STIM1 results in macroscopic CRAC currents exhibiting large inward currents with negligible outward currents at potentials more positive than ϩ60 mV ( Fig. 2A). In agreement with SOCE measurements, we observed currents in HEK293 cells expressing W176C-Orai1 even in the absence of STIM1, demonstrating the constitutive activity of these channels. Additionally, W176C mutant channels showed large outward currents in addition to inward currents, yielding an I/V with an S shape (Fig. 2B). Thus, in W176C mutant channels, currents are strongly suppressed up to ϩ40 mV, but in contrast to WT channels, large outward currents of monovalent Na ϩ or Cs ϩ ions develop at potentials more positive than ϩ50 mV ( Fig.  2B and supplemental Fig. 2A). These data suggest an important role of W176 in regulating the permeation properties of CRAC channels. Little is known about the molecular mechanism of CRAC current rectification, but it has been shown that removal of extracellular divalent ions is accompanied by loss of ion selectivity and permeation of Na ϩ and Cs ϩ both in the inward and outward direction (8). In this case, the I/V relationship of WT CRAC currents becomes more linear with symmetrical Na ϩ (see also Fig. 2C). Thus, extracellular Ca 2ϩ may play a role in limiting the outward current of monovalent cations at physiological Ca 2ϩ concentrations. Interestingly, the I/V relationship of W176C mutant Orai1 channels was completely linear in divalent free (DVF) solutions (Fig. 2D). Together these data suggest that large outward currents observed with W176C mutant channels may result from a change in CRAC channel pore affinity for extracellular Ca 2ϩ .
W176C Channels Show the Same K i for Ca 2ϩ Block of Monovalent Current as WT Channels-To further characterize the Ca 2ϩ -dependent blocking of inward and outward currents for WT and W176C mutant channels, we generated I/V relationships in the presence of various external Ca 2ϩ concentrations (Fig. 3, A and B). In the presence of increasing extracellular Ca 2ϩ , the inward monovalent currents decreased for both WT and W176C mutant channels. In some instances, the inward current amplitude for WT channels recovered at Ca 2ϩ concentrations greater than 6 mM. To quantify Ca 2ϩ blocking efficacy, we measured current amplitudes at fixed voltages (-110 mV and ϩ115 mV) for varying Ca 2ϩ concentrations. For inward currents, the plot of the current amplitude normalized to the maximum current amplitude in the absence of Ca 2ϩ as a function of extracellular Ca 2ϩ concentration is well fitted by a Hill equation with a coefficient approaching 1, suggesting a single binding site, and an IC 50 between 20 -40 M for both WT (Fig. 3A, center panel) and W176C (B, center panel) channels. This Ca 2ϩ affinity for channel blocking is similar to a previous report (8).
Regarding the Ca 2ϩ blocking efficacy of the Na ϩ inward current in W176C, it should be noted that at intermediary concentrations (100 M to 1 mM), the effect of Ca 2ϩ is time-and voltage-dependent (supplemental Fig. 2B). Thus, in contrast to WT channels where Ca 2ϩ blocking of the Na ϩ inward current is almost instantaneous, it takes tens of milliseconds to block the current in W176C. As a result, the affinity for Ca 2ϩ blocking measured at a steady state is similar to that observed in WT channels, but the value measured at the beginning during the peak current shifted 5-to 10-fold. Although extracellular Ca 2ϩ blocked WT and W176C inward currents with similar efficacy, blocking of the outward current was greatly different. In fact, WT outward currents were almost completely blocked by 1 mM Ca 2ϩ (IC 50 ϭ 255 M), but 100 mM Ca 2ϩ (IC 50 ϭ 6.2 mM) was required to achieve similar results with W176C mutant channels (Fig. 3, A and B, right panels).
We also examined how these outward currents developed and found further evidence in favor of Ca 2ϩ -dependent blocking of outward currents. In the absence of extracellular Ca 2ϩ , the outward current from W176C channels developed quasiinstantaneously (Fig. 3C, left panel) but developed slowly in the presence of external Ca 2ϩ , with a time constant of 2 to 5 ms depending on the voltage (Fig. 3C, right panel). These data sug-  OCTOBER 7, 2011 • VOLUME 286 • NUMBER 40 gest that the slow activation phase in the presence of Ca 2ϩ involves Ca 2ϩ unblocking. Because the rate of rise of the outward current is a function of voltage at the membrane potentials greater than the Ca 2ϩ reversal potential (Fig. 3D, left  panel), it follows that unblocking may be due to Ca 2ϩ dissociation from a site within the pore. As expected, the rate of unblocking was independent of the concentration of extracellular Ca 2ϩ (Fig. 3D, right panel).

Orai1TMIII Modulates CRAC Channel Gating and Permeation
So far, the W176C data uncover a blocking effect of Ca 2ϩ binding to a site within the membrane electrical field. In WT CRAC channels, extracellular Ca 2ϩ also blocks Na ϩ inward currents by binding at a site within the pore with an IC 50 close to 20 M (8). Because this IC 50 increases to 500 M in E106D mutant channels, which also show loss of ion selectivity, it has been suggested that this site was located at the selectivity filter (10). Interestingly, W176C mutant channels showed a very sim-

Orai1TMIII Modulates CRAC Channel Gating and Permeation
ilar IC 50 (20 M, Fig. 3A) for blockage of Na ϩ inward currents at hyperpolarized membrane potentials, suggesting that the binding site encompassing Glu-106 is not affected by the W176C mutation. However, our data also show that Ca 2ϩ efficacy at blocking Na ϩ outward currents is much lower in W176C mutant channels as compared with WT. As shown by our data, this loss in ion selectivity at the depolarized membrane potential is very likely due to Ca 2ϩ leaving its high-affinity binding site at Glu-106. A model is discussed later that includes Ca 2ϩ accumulation in the outer vestibule.
The Mutation of Gly-183 Causes STIM1-independent Channel Opening by Application of 2-APB-2-APB is widely used as a modulator of CRAC channel activity, which either activates or inhibits CRAC currents at concentrations of Ͻ 10 M and Ͼ 50 M, respectively (24). To examine whether currents observed with W176C and other mutants of Orai1 were modulated by 2-APB, we measured SOCE in the presence of 2-APB. Although W176C-mediated SOCE was unaffected by 2-APB (data not shown), one of the other mutants, G183A, which initially did not show significant Ca 2ϩ entry upon store depletion (Fig. 1B), showed a robust increase in cytoplasmic Ca 2ϩ in the presence of 2-APB (Fig. 4A). To examine whether this effect was depen-dent on store depletion, we exposed Orai1-deficient cells expressing W176C or G183A mutants to 2-APB in the absence of store-depletion (Fig. 4B). As expected, W176C showed elevated cytoplasmic Ca 2ϩ concentration, which remained unaffected by exposure to 2-APB (Fig. 4B). Interestingly, G183A showed enhanced cytoplasmic Ca 2ϩ upon exposure to 2-APB, even in the absence of store depletion (Fig. 4B). We then examined the permeation and gating properties of this mutant. Whole-cell patch clamp recordings showed that G183A mutant channels conducted Ca 2ϩ only in the presence of 2-APB. In this case, addition of 2-APB caused a rapid and simultaneous increase of inward and outward currents (Fig. 4C). It is unlikely that the development of outward and inward currents in G183A is due to separate processes because both effects occur

Orai1TMIII Modulates CRAC Channel Gating and Permeation
concomitantly. In WT channels, 2-APB activates outward currents, but this effect is delayed compared with the early activation and block of inward currents (supplemental Fig. 3,  A and B). Thus, in this respect, the channel permeation properties of G183A and W176C appear to be similar, both exhibiting large outward currents. The lack of G183A mutant channel activation by STIM1 suggests that this mutant may cause a collapse of the channel gate, which can be opened by 2-APB.
To further investigate the interaction of STIM1 with these Orai1 mutants, we measured W176C and G183A mutant channel currents in the presence and absence of STIM1. Both inward and outward currents were enhanced in W176C mutant channels by coexpression of STIM1. In contrast, currents remained unaltered by STIM1 in G183A mutant channels (Fig.  4D). In this case, the lack of effect of STIM1 is very likely due to the requirement of 2-APB to open the channel, which maximally activates CRAC channels, thereby masking the effect of STIM1.
Because Gly residues are considered helix breakers, we surmised that mutation in Gly-183 may uncouple the relay of the STIM1 signal for opening the gate at the selectivity filter region. To examine whether any other Gly residues within TMs of Orai1 function similarly, we systematically mutated all the Gly residues within TMs of Orai1 (supplemental Fig. 4). Orai1 encodes three glycine residues, one in TM1, a second in TMIII, and the third one in TMIV. All the Gly mutants showed a similar expression as judged by immunoblotting and immunocytochemistry ( Fig. 1C and supplemental Fig. 4B). Mutation of Gly in TM1 (G98A) showed a very low SOCE that was blocked instantaneously and remained unaffected by external application of 2-APB (supplemental Fig. 4C). Accordingly, we could not detect any CRAC currents by expression of this mutant in HEK293 cells in the presence or absence of STIM1 (data not shown). On the other hand, Gly in TMIV (G247A) showed increased SOCE (supplemental Fig. 4C) but other than that behaved exactly like WT Orai1 in terms of permeation and gating characteristics (supplemental Fig. 4D and data not shown). In summary, we observed STIM1-independent Ca 2ϩ entry in the presence of 2-APB only with G183A mutant, indicating a specific role of TMIII.
The W176C Mutant Localizes and Translocates Normally under Resting and Store-depleted Conditions-It has been shown that interaction of STIM1 with Orai regulates at least two separate processes: one responsible for clustering of Orai1 into the junctional regions between the PM and ER membrane and the other required for channel activation. Because both W176C and G183A mutants show STIM1-independent currents, we examined whether they exhibited proper clustering with STIM1. To this end, we measured their translocation to PM/ER junctional regions upon stimulation with thapsigargin using TIRF microscopy. As reported previously (22), WT Orai1 channels are homogenously distributed in the PM, whereas-STIM1 is distributed in the ER membrane and hence shows a very low signal under resting conditions (Fig. 5A, top panels). Upon store depletion, however, STIM1 clusters and translocates to near PM junctional areas, and Orai1 accumulates into those clusters (Fig. 5A, lower panels). Under the same condition, both W176C and G183A mutant channels clustered into junctional areas with kinetics similar to that of WT channels (Fig. 5, B and C, and data not shown). These results suggest that the W176C and G183A mutants retain at least the interaction with STIM1 required for their clustering in junctional areas. Thus, despite their clustering into junctional areas upon coexpression with STIM1 (Fig. 5), the G183A mutant channels remained closed upon store depletion and were activated only by extracellular 2-APB. This observation supports the hypothesis that TMIII plays a role in transmitting a STIM1-dependent activation signal from the cell interior to the upper channel gate, and G183A alters this process.
The W176C Mutant and D110A/D112A Have Similar Effects on CRAC Channel Permeation Properties-It has been reported that the Asp-110A/Asp-112A mutant Orai1 channels showed a Ca 2ϩ affinity similar to WT despite the presence of outward currents at depolarized membrane potentials (13). On the basis of the similarity between the effects of W176C and D110A/ D112A on CRAC channel permeation properties, we surmised that the W176C mutation may affect the positioning of Asp-110/Asp-112 on the outside of the channel pore. To test this hypothesis, we measured CRAC currents generated by the double mutant channel D110A/D112A coexpressed with STIM1 in HEK293T cells. With this double mutant, we observed large outward currents that develop with a slow time course (Fig. 6A), suggesting Ca 2ϩ unblocking at depolarized membrane potentials. Also, as with W176C mutant channels, high concentrations of extracellular Ca 2ϩ (100 mM) were necessary to block these outward currents (Fig. 6, B-D), and this block occurs very slowly in minutes, supporting the hypothesis that the double D110A/D112A mutation prevents Ca 2ϩ accumulation in the outer vestibule. These data, together with that of others showing that Na ϩ inward currents generated by the double mutation are blocked by Ca 2ϩ with high affinity (13), demonstrate that the W176C and D110A/D112A mutations similarly affect CRAC channel permeation properties.

DISCUSSION
The key properties of CRAC currents include a very high selectivity for Ca 2ϩ over monovalent ions and inward rectification. Our data show that mutation of the Trp-176 and Gly-183 residues (W176C and G183A) in Orai1 TMIII exhibited strong outward currents. In addition, W176C was constitutively active in the absence of STIM1. On the other hand, G183A mutant channels were closed even when coexpressed with STIM1, but addition of 2-APB activated the channel irrespective of STIM1. These data suggest an important role of TMIII in CRAC channel gating, although it has been shown to not line the pore (25,26).
In cation-selective channels, inward rectification may result from two processes: concentration of ions at the outer mouth of the channel by pore-associated negative charges, which facilitates inward currents (27), or blocking of the outward current as seen in inward-rectifying K ϩ channels. The latter may result from Mg 2ϩ and polyamines entering the channel (28,29), asymmetry of the ion conduction pathway, or positive charges located at the inner mouth of the pore that repel cations (30). Inward rectification in CRAC channels may involve both, facilitation of inward currents and blocking of outward currents. Our data obtained with W176C and G183A indicate that blocking of the outward current by Ca 2ϩ plays a role in the rectification process. The I/V of W176C and G183A is characterized by strong current suppression up to ϩ40 mV and large outward currents above ϩ50 mV. This pattern is abolished in the absence of extracellular Ca 2ϩ , giving rise to a linear I/V. We suggest that current inhibition up to ϩ40 mV is due to Ca 2ϩ blocking on the basis of a number of observations. The outward currents in W176C and G183A mutant channels develop slowly in the presence of Ca 2ϩ , with a time constant of 2-5 ms, whereas activation is instantaneous in the absence of Ca 2ϩ (Fig.  3 and data not shown). The rate of rise is strongly voltage-dependent, increasing with positive membrane potentials, and independent of extracellular Ca 2ϩ concentration. This suggests that the Ca 2ϩ blocking site lies within the membrane electrical field. Furthermore, the K off (rate of Ca 2ϩ from its blocking site) derived from the rate of activation of outward currents are approximately 650 s Ϫ1 at 115 mV and 300 s Ϫ1 at 60 mV, which is consistent with a K off of 125 s Ϫ1 estimated for the blocking effect of Ca 2ϩ on inward currents at Ϫ100 mV (10). Finally, the IC 50 for Ca 2ϩ , close to 20 M, is almost identical to the value estimated for Ca 2ϩ -dependent blocking of the Na ϩ inward current (8). On the basis of the observation that Ca 2ϩ blocking of Na ϩ inward currents in WT CRAC channels decreased with the mutation of Glu-106 (E106D), the IC 50 shifting from 20 M to 500 M, it has been suggested that the Ca 2ϩ blocking binding site was at the selectivity filter. In addition, a Hill coefficient of 1 deduced from the blocking effect of Ca 2ϩ in WT and W176C mutant channels indicates that Ca 2ϩ binds to a single site (Glu-106) that remains unaltered in the W176C mutant. Because this site is functional in W176C mutant channels as judged by a Ca 2ϩ affinity identical to WT channels for inward currents, the observation of outward currents at depolarized potentials in W176C channels is puzzling. In fact, these observations parallel those with D110A/D112A mutant Orai1 channels, showing a Ca 2ϩ affinity similar to WT and the presence of outward currents at depolarized membrane potentials ( Fig. 6 and Ref. 13). On the basis of these observations, we propose a model for Orai (Fig. 7) derived from a BK Ca channel model (31). In Orai1, the Glu-106 residue forms the sole Ca 2ϩ binding site that controls ion selectivity and the Asp-110 and Asp-112 residues, which also line the pore but above the selectivity filter allow for local Ca 2ϩ concentration in the outer vestibule. Such Ca 2ϩ accumulation has two major effects: increase Ca 2ϩ inward current via a mechanism similar to that proposed for BK Ca channels (31) and retention of Ca 2ϩ at the Glu-106 binding site at depolarized membrane potentials. The presence of Ca 2ϩ at this site prevents the loss of ion selectivity and generation of outward currents of monovalent cations at depolarized membrane potentials. Mutation of Asp-110 and Asp-112, on the other hand, dissipates the high Ca 2ϩ ion density in the outer vestibule. Under this condition, the affinity of the Glu-106 binding site for Ca 2ϩ is not affected, but association of Ca 2ϩ with the site may be slower at low concentrations. The major effect of this mutation would be to promote the displacement of Ca 2ϩ from the Glu-106 binding site at depolarized membrane potential and account for the loss of ion selectivity and generation of outward currents. Our data indicate that the mutations in TMIII do not affect Ca 2ϩ binding to the Glu-106 site. However, it may affect the mechanism of Ca 2ϩ accumulation in the outer vestibule. Although mutation of Glu-190 in TMIII causes a similar activation of the Na ϩ outward current (supplemental Fig. 5), it is unlikely to directly affect the negative charge distribution in the outer mouth of the channel because TMIII does not line the pore (25,26). It follows, therefore, that the effects of TMIII on the ion selectivity at depolarized membrane potentials is likely mediated via allosteric rearrangement of Asp-110 and Asp-112 in the outer vestibule (Fig. 7, lower panel). Whether Glu-190 or the TMIII/TMIV extracellular loop play a role in this process remains to be demonstrated.
As reported for many Ca 2ϩ and K ϩ channels (32-34) CRAC channel gating may occur at the selectivity filter (10,11). Accordingly, our W176C and G183A data may be explained assuming a gate located at the outer mouth of the channel near the selectivity filter. In WT channels, a Trp-176-dependent process maintains the channel gate in a closed state in the absence of STIM1. TMIII may transmit this Trp-176-dependent closing signal to the gate. STIM1, but also the mutations W176C and G183A, release the closing mechanism, allowing the channel to open. Thus, the W176C and G183A mutants channel open in a STIM-independent fashion. Whether the Orai channel gate involves the selectivity filter itself or a structure near it is unclear, but it is also possible that the gate is above the selectivity filter and comprises the negatively charged residues Asp-110 and Asp-112. Such a mechanism has been postulated to account for Kv4.3 gating (34). Finally, the CRAC channel gate may be located at the intracellular face of the pore, and the large side chain of Trp-176 may regulate this mechanism. There is no direct evidence in favor of this hypothesis, but it should be noted that the mutations R91W and K85A/K87A (Ref. 35 and data not shown) at the base of TMI block channel permeation, suggesting an important role of this region in channel gating.
In summary, our data obtained with the W176C and G183A mutant channels indicate that TMIII regulates both permeation and gating of CRAC channels. Allosteric interaction of TMIII with the outer mouth of the pore may prevent activation of monovalent cation outward currents. This process does not directly involve the high-affinity binding site for Ca 2ϩ in the selectivity filter at Glu-106 but rather some negative charges that cause local concentration of Ca 2ϩ in the outer vestibule. This local accumulation of Ca 2ϩ maintains Ca 2ϩ binding to Glu-106 and thus prevents monovalent cation outward currents at depolarized membrane potentials. Finally, our data indicate that Glu-176 is part of a mechanism that maintains Orai in a closed state. STIM frees this mechanism, and the opening signal is transmitted to the gate via TMIII.