Stoichiometric requirements for trapping and gating of Ca²⁺ release-activated Ca²⁺ (CRAC) channels by stromal interaction molecule 1 (STIM1)

Abstract

Store-operated Ca²⁺ entry depends critically on physical interactions of the endoplasmic reticulum (ER) Ca²⁺ sensor stromal interaction molecule 1 (STIM1) and the Ca²⁺ release-activated Ca²⁺ (CRAC) channel protein Orai1. Recent studies support a diffusion-trap mechanism in which ER Ca²⁺ depletion causes STIM1 to accumulate at ER-plasma membrane (PM) junctions, where it binds to Orai1, trapping and activating mobile CRAC channels in the overlying PM. To determine the stoichiometric requirements for CRAC channel trapping and activation, we expressed mCherry-STIM1 and Orai1-GFP at varying ratios in HEK cells and quantified CRAC current (I_CRAC) activation and the STIM1:Orai1 ratio at ER-PM junctions after store depletion. By competing for a limited amount of STIM1, high levels of Orai1 reduced the junctional STIM1:Orai1 ratio to a lower limit of 0.3–0.6, indicating that binding of one to two STIM1s is sufficient to immobilize the tetrameric CRAC channel at ER-PM junctions. In cells expressing a constant amount of STIM1, CRAC current was a highly nonlinear bell-shaped function of Orai1 expression and the minimum stoichiometry for channel trapping failed to evoke significant activation. Peak current occurred at a ratio of ∼2 STIM1:Orai1, suggesting that maximal CRAC channel activity requires binding of eight STIM1s to each channel. Further increases in Orai1 caused channel activity and fast Ca²⁺-dependent inactivation to decline in parallel. The data are well described by a model in which STIM1 binds to Orai1 with negative cooperativity and channels open with positive cooperativity as a result of stabilization of the open state by STIM1.

Store-operated Ca²⁺ entry (SOCE) is a widely distributed process critical for the function of the immune system and a variety of excitable and nonexcitable tissues (1–3). SOCE is evoked by plasma membrane (PM) receptors that stimulate inositol 1,4,5-trisphosphate–dependent release of Ca²⁺ from the endoplasmic reticulum (ER); the ensuing depletion of ER Ca²⁺ opens store-operated channels in the PM, the best characterized of which is the Ca²⁺ release-activated Ca²⁺ (CRAC) channel (4). Recent studies have identified stromal interaction molecule 1 (STIM1) as the ER Ca²⁺ sensor for SOCE (5–7) and Orai1 as the pore-forming subunit of the CRAC channel (reviewed in 1). A tetramer of Orai1 subunits is thought to form the active CRAC channel (8, 9).

A currently accepted mechanism for SOCE is that ER store depletion causes STIM1 to oligomerize (10, 11), driving the redistribution of STIM1 and Orai1 to sites of close apposition between the ER and the PM known as ER-PM junctions (5, 12, 13). The colocalized proteins at junctions appear as “puncta” at the light microscope level in cells expressing fluorescently labeled STIM1 and Orai1. Several observations support a two-part diffusion trap mechanism for the redistribution of STIM1 and Orai1 to ER-PM junctions. After store depletion, STIM1 forms puncta independently of Orai1, possibly via interactions of its C-terminal polybasic domain with negatively charged phosphoinositides in the PM (refs. 11, 14–16; but see ref. 17). Junctional STIM1 traps and activates diffusing CRAC channels (18, 19) through the binding of Orai1 to the STIM1 CRAC activation domain [CAD, amino acids 342–448 (15), also called the STIM1-Orai activating region (SOAR) (20)]. CAD interacts with the N- and C-terminal domains of Orai1; binding to the C terminus is stronger and is necessary and sufficient for Orai1 clustering at ER-PM junctions, whereas interactions with the N terminus are weaker but required for channel opening (15, 21). CRAC channels also inactivate in response to binding of incoming Ca²⁺ to sites near the channel pore (22, 23), and STIM1 plays a critical role in promoting this process (24–26).

Two recent studies suggest that the STIM1:Orai1 binding stoichiometry can affect the gating behavior of CRAC channels. Scrimgeour et al. (27) found that the relative abundance of STIM1 and Orai1 determined the extent of CRAC current (I_CRAC) inactivation, and Li et al. (28) reported that Ca²⁺ currents increased with the number of STIM1 binding sites on engineered tetrameric Orai1 concatemers. However, the stoichiometric requirements for the trapping and gating of native CRAC channels are unknown. In this study, we developed methods to control and quantitate the ratio of STIM1 and Orai1 at ER-PM junctions while monitoring CRAC channel behavior. We found that activation and inactivation of CRAC channels are highly nonlinear functions of the STIM1:Orai1 ratio; although binding of one to two STIM1s is enough to trap CRAC channels, eight STIM1s must bind for full activation and inactivation.

Results

For simplicity, we refer to STIM1 and Orai1 throughout as STIM and Orai, respectively, and to ER-PM junctions as puncta. Our basic strategy is schematized in Fig. 1A; when STIM is present in excess over Orai, it is expected to saturate the binding sites on each CRAC channel, whereas increased expression of Orai will deplete the pool of free STIM, reaching a limit at which the channels in puncta will each be bound to the minimum number of STIMs necessary for immobilization.

Fig. 1.

Minimum STIM:Orai stoichiometry for CRAC channel trapping at ER-PM junctions. (A) Experimental design. At low Orai expression, STIM is in excess and saturates CRAC channel binding sites at ER-PM junctions after store depletion (Upper); when Orai is in excess, it reduces the number of STIMs bound per channel to the minimum necessary for trapping (Lower). (B) Confocal images of the footprint of HEK 293 cells expressing mCh-GFP calibrator or Orai-GFP and mCh-STIM. After store depletion with 1 μM TG, Orai and STIM redistribute into colocalized puncta but mCh-GFP calibrator maintains a reticular pattern. (C) Pseudocolor image of mCh:GFP ratios in the cells from B. The STIM:Orai ratio in puncta is 0.48 ± 0.14 (mean ± SD) relative to the mCh:GFP ratio of 1.0 ± 0.38 (mean ± SD) in the calibrator cells. (D) STIM:Orai ratios in puncta (means ± SD) from single cells as a function of the mean peripheral Orai:STIM ratio measured at the cell footprint. mCh-STIM expression was allowed to vary (red dots) or was held constant (0.21 ± 0.01 a.u.; black dots). The maximum estimated error from free Orai background is indicated by the open symbols.

Quantifying the STIM:Orai Ratio in Puncta.

HEK 293 cells stably expressing mCherry (mCh)-STIM were transiently transfected with Orai-GFP and imaged by confocal microscopy at the cell footprint. After depletion of ER Ca²⁺ stores by thapsigargin (TG), STIM and Orai redistributed from diffuse locations in the ER and PM into colocalized puncta at ER-PM junctions (Fig. 1B). To convert mCh:GFP fluorescence ratios to STIM:Orai ratios, we designed a calibrator protein containing mCh and GFP in a 1:1 ratio. The calibrator was constructed from STIM by deleting residues 237–427 and adding mCh and GFP to the N and C termini, respectively. This design exposes mCh and GFP labels to the same environments (ER and cytosol, respectively) as the mCh-STIM and Orai-GFP labels, avoiding possible distortion of ratios from environmental effects. Also, the distance between the mCh and GFP on the calibrator prevents intramolecular FRET, which would otherwise distort the fluorescence ratio (Fig. S1). Finally, the deletion of cytosolic coiled-coil domains prevents STIM redistribution after store depletion (29, 30), making it possible to distinguish cells transfected with the calibrator alone from those expressing mCh-STIM and Orai-GFP in the same field of view (Fig. 1B). The calibrator fluorescence ratio was relatively uniform within and among cells (Fig. 1C), indicating a consistent mCh:GFP stoichiometry. We applied a covariance-based method (15) to mask regions of colocalized mCh-STIM and Orai-GFP puncta in store-depleted cells; for each of these regions, the mCh:GFP fluorescence ratio was divided by the mean mCh:GFP calibrator ratio to yield the STIM:Orai ratio (Fig. 1C).

The Limiting STIM:Orai Stoichiometry for CRAC Channel Trapping.

To produce different ratios of STIM and Orai in puncta, we varied their relative levels of expression in HEK cells. Protein expression was quantified in single cells from confocal fluorescence images of the cell footprint after store depletion; the resulting peripheral Orai:STIM ratios were then used to indicate the relative abundance of both proteins in equilibrium with bound complexes in puncta. In cells expressing a similar amount of STIM, increasing levels of Orai diminished the STIM:Orai ratio measured in puncta (Fig. S2). To determine the minimum STIM:Orai ratio sufficient for channel trapping, we analyzed cells expressing variable levels of both proteins or cells with variable Orai-GFP but a fixed level of mCh-STIM (0.21 ± 0.01 a.u., n = 15) (Methods). In both sets of experiments, as Orai expression was increased relative to STIM, the STIM:Orai ratio in puncta reached a minimum plateau of ∼0.3 (Fig. 1D). The fact that the relative amounts of Orai and STIM, rather than the absolute Orai level, determined the minimum plateau value implies that the plateau results from the competition of Orai for a limited pool of STIM rather than for a potential third protein involved in complex formation (e.g., CRACR2A) (31).

We considered how background fluorescence from STIM or Orai could affect these puncta measurements. Background from STIM was not likely to be a problem, because bulk ER fluorescence intensity of mCh-STIM was typically <10% of that in nearby puncta (Fig. 1B and Fig. S2 A and C). However, excess Orai could occupy free space in the puncta to reduce the apparent STIM:Orai binding ratio, especially at the highest expression levels. An upper limit for this error was estimated by assuming that the entire puncta area was available for free Orai diffusion and subtracting the fluorescence of unbound Orai outside puncta from the puncta fluorescence. This correction increased the estimated limiting STIM:Orai ratio in puncta to ∼0.6 (Fig. 1D, open symbols). Thus, we conclude that binding of one to two STIMs is sufficient to immobilize a tetrameric CRAC channel at the ER-PM junction.

CRAC Channel Activation Is a Steep Function of STIM:Orai Stoichiometry.

Key questions are whether this minimal trapping stoichiometry also activates channels, and, more generally, how CRAC channel activity depends on the number of bound STIMs. To address this, we measured I_CRAC amplitude as a function of Orai expression in individual HEK cells expressing approximately the same level of STIM. After I_CRAC was induced to a steady-state level by 10 mM EGTA in the pipette, the current density (current normalized to cell capacitance) at −100 mV was measured and plotted against the cell's density of Orai-GFP expression as estimated from the PM-associated GFP fluorescence (Fig. 2A and Fig. S3). In the low range of Orai expression, I_CRAC increased with Orai as expected but dropped abruptly above an apparent Orai expression threshold of ∼1 a.u. The current-voltage relation remained constant throughout the entire range of Orai expression (Fig. S4), showing inward rectification characteristic of native I_CRAC. In cells with high Orai expression and little or no I_CRAC, Orai and STIM were colocalized in puncta, indicating that the lack of current was not attributable to a failure of STIM and Orai to interact (Fig. S4). Thus, binding of one to two STIMs, although sufficient to trap channels in puncta, does not activate them significantly.

Fig. 2.

CRAC channel activity is a highly nonlinear function of Orai expression. (A) I_CRAC as a function of Orai-GFP expression in single cells. Each point is the current evoked by a brief hyperpolarization to −100 mV, normalized for cell capacitance (Methods). Cells expressed roughly the same level of mCh-STIM (0.21 ± 0.01 a.u., n = 37 cells) and varying amounts of Orai-GFP. In the range of Orai expression where I_CRAC declined sharply, fluorescence measurements in a subset of cells (open symbols) indicated a peripheral Orai:STIM ratio of 0.55 ± 0.02 (mean ± SEM), corresponding to 1.82 ± 0.06 STIMs per Orai subunit. (B) Reduced level of STIM lowers the threshold for Orai-induced suppression of SOCE. After store depletion, Ca²⁺ influx rates [d(F₃₅₀/F₃₈₀)/dt] were measured on Ca²⁺ readdition (Fig. S5) in cells with moderate (0.22 ± 0.01 a.u., n = 68; circles) or low (0.10 ± 0.01 a.u., n = 48; triangles) levels of mCh-STIM. Each point is the mean ± SEM of the Ca²⁺ influx rate for 4–12 cells, averaged in bins of 0.2 a.u. of Orai expression.

High Orai levels may suppress I_CRAC by depleting the pool of free STIM, and thereby reducing the STIM:Orai binding stoichiometry, or by depleting the supply of another binding protein, making it unavailable to participate in CRAC channel activation. To distinguish between these interpretations, we measured the effect of Orai expression on CRAC channel activity in cells expressing different levels of STIM. CRAC activity was assessed in fura-2–loaded cells by depleting stores under Ca²⁺-free conditions and measuring the rate of Ca²⁺ entry upon Ca²⁺ readdition (Fig. S5). In cells expressing a moderate level of mCh-STIM (0.22 ± 0.01 a.u., n = 68 cells), the influx rate was a bell-shaped function of Orai expression (Fig. 2B), similar to the I_CRAC vs. Orai relation from cells expressing a comparable amount of STIM (compare with Fig. 2A). However, in cells expressing about 50% less mCh-STIM (0.10 ± 0.01 a.u., n = 48 cells) the suppression of Ca²⁺ entry rate was shifted to lower levels of Orai. Such a shift is not expected from depletion of a third component by Orai, but instead supports the idea that excess Orai suppresses CRAC channel activity by reducing the STIM:Orai binding stoichiometry.

Thus, the steep decline of CRAC current at Orai levels of 1–2 a.u. (Fig. 2A) implies that activation is a highly nonlinear function of STIM:Orai binding, falling abruptly at stoichiometries below the level required for full activity. In six cells in this range of Orai (Fig. 2A, open symbols), the peripheral STIM:Orai ratio was 1.82 ± 0.06 (mean ± SEM; complete dataset in Fig. S6). These results suggest that binding of two STIMs per Orai, or eight STIMs per channel, is required to activate the CRAC channel fully.

STIM:Orai Stoichiometry Affects Rapid Ca²⁺-Dependent Inactivation.

Ca²⁺-dependent inactivation (CDI) was measured from the decay of I_CRAC during brief hyperpolarizations. At low Orai levels, CDI was prominent (Fig. 3A, traces 1 and 2), but it declined with increasing Orai expression and was replaced by potentiation at the highest Orai levels (traces 3–5; summarized in Fig. 3B). To relate these changes to the degree of channel activation, the I_CRAC vs. Orai expression data are plotted in Fig. 3C using pseudocolor to encode the amount of inactivation or potentiation. All cells with less than 1 a.u. of Orai displayed the normal extent of CDI. In the region where I_CRAC declined abruptly with increases in Orai (1–2 a.u.), cells with currents larger than −25 pA/pF displayed normal CDI, whereas those with smaller currents showed reduced CDI or slight potentiation. Finally, cells at the highest Orai levels (>3 a.u.) had small currents that strongly potentiated. These altered current kinetics cannot be explained by a reduced number of fully liganded channels; instead, they most likely arise from channels complexed with suboptimal numbers of STIMs. These data demonstrate that both activation and inactivation of CRAC channels are determined in a similarly nonlinear way by the STIM:Orai stoichiometry.

Fig. 3.

Rapid CDI is a highly nonlinear function of Orai expression. Data are from the cells of Fig. 2A. (A) Inactivation during hyperpolarizing steps is progressively replaced by potentiation in cells expressing increasing amounts of Orai. Each trace is the response to a step from +30 mV to −120 mV in 20 mM Ca²⁺. (B) Proportion of current remaining at the end of the hyperpolarization, quantified as the current at 195 ms (I₁₉₅) relative to that at 3 ms (I₃), as a function of Orai expression in single cells. (C) At high levels of Orai, the loss of CDI and appearance of potentiation occur in parallel with the increasing suppression of CRAC channel activity. The data in Fig. 2A are reproduced with colors indicating the current remaining at the end of the pulse.

A Model to Describe CRAC Channel Activation by STIM.

To gain further insight into CRAC channel gating, we fitted a modified Monod–Wyman–Changeux model to the I_CRAC vs. Orai relation (SI Methods). The model assumes two channel conformations (open and closed), each with four binding sites for STIM, and gating cooperativity arises from the incremental stabilization of the open state by each bound STIM ligand. No assumption is made about the stoichiometry of the STIM ligand (e.g., whether 1 or 2 STIMs bind to each site). The model has five parameters: the opening equilibrium constant, L; the stabilization factor, f; the total STIM ligand concentration, S_total; the association constant, K_a; and a binding cooperativity factor, a. The choices for L and f were constrained by the open probability (P_o) of CRAC channels at rest (∼0) and after activation by store depletion (∼0.8) (32). The only free parameters are S_total, K_a, and a, which were adjusted to provide the best fit to the peak and steepness of the I_CRAC vs. Orai relation (SI Methods).

With the assumption that binding to each site is independent and noncooperative (a = 1), the model cannot account for both the required steepness of the I_CRAC vs. Orai relation and the predominance of partially occupied states that underlie noninactivating currents at high Orai levels (Fig. S7). However, this shortcoming was overcome by assuming that STIM binds to Orai with negative cooperativity (a = 0.5). With this assumption, the model provides a reasonable fit to the I_CRAC vs. Orai data in Fig. 4B and predicts that the open states OS₂ and OS₃ will dominate the current in the Orai range of 3–7 a.u. (Fig. 4 D and E). The model also accurately predicts the relation between STIM:Orai ratios in puncta and the peripheral Orai:STIM expression ratio, given the assumption that binding of one or two STIMs to a single site is sufficient to trap channels (Fig. 4C).

Fig. 4.

Equilibrium model for CRAC channel gating by STIM. (A) Modified Monod–Wyman–Changeux scheme in which closed (Left) and open (Right) channel states each have four STIM binding sites, with bound sites indicated in black. Equilibrium constants for each transition are shown. K_a, STIM association constant; L, opening equilibrium constant; f, opening cooperativity factor; a, binding cooperativity factor. (B) Best fit of the model to the I_CRAC vs. Orai data of Fig. 2A, with K_a = 100, L = 10⁻⁴, f = 14.2, a = 0.5, and S_total = 3.2. (C) Model predictions superimposed on the data of Fig. 1D, assuming that channel trapping in puncta requires binding of one (blue) or two (green) STIMs to a single site. (D) Predictions of open-channel state occupancies and free STIM concentration as a function of total Orai expression. (E) Composition of the open-channel population as a function of Orai expression. In the range of moderate Orai (3–7 a.u.) in which CDI transitions to potentiation (Fig. 3C), OS₃ and OS₂ dominate over OS₄.

Discussion

In this study, we combined whole-cell recording with single-cell measurements of STIM and Orai to determine the stoichiometric requirements for CRAC channel trapping and gating at ER-PM junctions. At high levels of Orai expression relative to STIM, the limiting STIM:Orai ratio in puncta revealed that one to two STIMs per tetrameric CRAC channel are sufficient for immobilization. Given evidence discussed below that a dimer of STIM binds to each of four sites on the CRAC channel, we believe that the minimal requirement for trapping is likely to be two STIMs binding to one site on the tetrameric channel. Alternatively, under these limiting conditions, a dimer of STIM may link two Orai dimers, as proposed by Penna et al. (9). In either case, the minimal binding stoichiometry for trapping was not sufficient to generate significant channel activity. In some prior studies, heterologous expression of Orai suppressed endogenous SOCE or I_CRAC (33–36), prompting the notion that STIM binding to multiple binding sites may be needed to trigger channel opening (35). Our results provide quantitative support for this idea by showing that channels with one to two STIMs bound are inactive.

CRAC channel activity is a highly nonlinear function of STIM binding, falling sharply at STIM:Orai ratios below ∼2, or eight STIMs per channel. Our model predicts that this transition occurs when fewer than four sites are occupied, suggesting that binding of one STIM dimer to each of four sites is necessary to confer maximal activity. These results are mostly in agreement with a recent study by Li et al. (28) of tetrameric concatemers of Orai and Orai tethered to dimers of a STIM cytoplasmic domain. Based on the maximal currents produced by four tethered STIM dimers per channel, the authors concluded that STIM binds to each site on the channel as a dimer. Using tetrameric Orai concatemers with variable numbers of STIM binding sites, Li et al. (28) also showed that current increased with STIM binding, but in a much less nonlinear way than we found using monomeric Orai. Our model predicts that channels with one, two, or three sites bound have open probabilities of 0.001, 0.025, and 0.275, respectively, relative to the fully occupied channel, compared with ∼0.07, 0.21, and 0.54 in the study by Li et al. (28). The basis for these differences is not clear, but studies of concatemerized voltage-gated K⁺ (K_V) channels and GABA receptors have revealed abnormal gating kinetics and ligand binding affinity that were attributed to intersubunit linkers (37, 38). It will be important to test whether the linkers introduced in Orai concatemers reduce the overall cooperativity of gating or whether an excess of free Orai in our experiments might accentuate nonlinear activation behavior, perhaps through contact-mediated inhibition of STIM-bound channels.

As Orai levels surpassed a threshold, CDI was lost (and potentiation appeared) in parallel with the decline in channel activation. These results confirm and extend work by Rychkov and colleagues (27), who reported that the extent of CDI varied with the transfection ratio of STIM and Orai cDNAs. The authors concluded that a high STIM:Orai binding stoichiometry favors CDI, whereas a low stoichiometry promotes potentiation during hyperpolarizing pulses. We found that CDI was maximal at Orai levels that produced maximal channel activation; thus, we hypothesize that like activation, maximal inactivation requires binding of eight STIMs and that as channels engage fewer STIMs, they shift from inactivation to potentiation. Because OS₁ is largely inactive, it is most likely that OS₂ and/or OS₃ underlies the currents with altered kinetics. Interestingly, the constant shape of the current-voltage relation among cells exhibiting CDI, no CDI, or potentiation (Fig. S4) implies that the number of STIMs bound to the channel does not grossly affect selectivity. Exogenous Orai displayed strong inactivation similar to that of endogenous CRAC channels only under conditions where it was occupied by at least eight STIMs per channel. This result implies that after store depletion, active endogenous CRAC channels are maximally bound by STIM, consistent with the high P_o (0.8) of endogenous CRAC channels measured in store-depleted Jurkat T cells (32).

The mechanism of CDI is complex, with recent studies revealing roles for a calmodulin binding domain in the Orai N terminus (24), the intracellular II–III loop of Orai (39), and an acidic cytosolic domain of STIM (ID_STIM; amino acids 470–491) (24–26). The requirement for ID_STIM in CDI offers several potential explanations for why CDI increases with the STIM:Orai stoichiometry. First, the extent of CDI may be determined by the number of ID_STIM domains that physically interact with the channel complex, in much the same way as activation depends on the number of bound STIMs (or, more precisely, CADs). A second possibility is that STIM influences the extent of CDI by determining the channel P_o; because STIM binding stabilizes the open state, the inactivation machinery is exposed to high intracellular Ca²⁺ ([Ca²⁺]_i) for a greater fraction of time, thereby promoting CDI. Local [Ca²⁺]_i within the ER-PM junction may also increase with P_o to enhance CDI.

To account adequately for the I_CRAC vs. Orai relation and the predominance of partially liganded channels at high Orai levels, we incorporated elements of both positive and negative cooperativity in our model. Positive cooperativity of the opening transition in the model results from stabilization of the open state by STIM. Negative cooperativity of STIM binding implies that each successive binding event is progressively less energetically favorable; this could arise from several sources, including allosteric effects on Orai or conformational or steric constraints associated with binding eight STIMs within the confines of a single channel. Of course, although the model is simple and highly constrained, it is certainly not unique, and direct assays will be needed to test its predictions, including the negative cooperativity of binding.

The highly nonlinear dependence of CRAC channel activation on STIM:Orai stoichiometry has interesting implications for function under physiological conditions. After Ca²⁺ is released from the ER, the time it takes for STIM at ER-PM junctions to reach “threshold” for fully activating CRAC channels may add significantly to the delay between store depletion and Ca²⁺ entry (e.g., ref. 12). This delay is thought to be a major determinant of Ca²⁺ oscillations in T cells (40), which may, in turn, enhance the efficiency and specificity of Ca²⁺ transcription coupling (41). Once the STIM threshold is reached, individual channels would be expected to open abruptly, consistent with results of noise analysis suggesting that the slow increase in whole-cell I_CRAC following store depletion results from the stepwise recruitment of closed channels to a high-P_o state (32). In this way, nonlinear activation of CRAC channels by STIM may give rise to complex history-dependent Ca²⁺ signaling dynamics that depend on the relative locations and abundance of STIM and Orai.

Methods

Cells and Constructs.

HEK 293 cell lines containing an inducible mCh-STIM or mCh-GFP calibrator (SI Methods) were generated with the Flp-In T-REx system (Invitrogen) and cultured as described previously (15). The mCh-GFP calibrator construct was made by removing amino acids 237–427 from mCh-STIM (18) and ligating it into a GFP-N1 vector.

Microscopy.

Confocal microscopy was carried out at 22–25 °C on a Leica SP2 AOBS confocal microscope system as described (15). Wide-field epifluorescence microscopy was performed at 22–25 °C with an Axiovert 200M microscope equipped with a Fluar 40× 1.3-N.A. oil immersion objective (Carl Zeiss MicroImaging), a monochromator light source (Polychrome II; Till Photonics), and a cooled CCD camera (ORCA-ER; Hamamatsu) with 2 × 2 pixel binning (12).

In Situ Fluorescence Measurements of STIM and Orai.

mCh-STIM and Orai-GFP puncta were identified as areas of high fluorescence colocalization after store depletion, as described previously (15). High overexpression of STIM and Orai can generate puncta in resting cells; we restricted our analysis to cells in which puncta appeared only after store depletion. The mCh:GFP ratios were calculated for individual puncta across the footprint using an automated particle analysis routine in ImageJ (W. S. Rasband, National Institutes of Health, Bethesda, MD; http://rsb.info.nih.gov/ij/) and then divided by the mean mCh:GFP calibrator fluorescence ratio to yield STIM:Orai ratios.

The Orai:STIM expression ratio was measured in the cell periphery after store depletion. In confocal experiments, fluorescence was integrated across the cell footprint, excluding peripheral lamellae, and in electrophysiology experiments, fluorescence was integrated in a 1-μm band around the cell's edge in cell equator images. The calibrator was used to convert mCh:GFP ratios to STIM:Orai ratios.

Levels of Orai-GFP and mCh-STIM expression were assessed in resting cells by normalizing fluorescence to solution standards. In confocal experiments, mean fluorescence across the cell footprint was normalized to bath-applied fluorescein (6.5 μM; Sigma) or cadaverine 594 (360 pM; Invitrogen). In wide-field experiments (electrophysiology and Ca²⁺ imaging), mean mCh-STIM fluorescence within 1 μm of the PM and the mean Orai-GFP fluorescence peak from eight radial intensity profiles intersecting the PM were measured at the cell equator and normalized to fluorescein (100 nM; Sigma) or rhodamine (2.5 μM; Sigma) solutions. All images were analyzed with ImageJ (National Institutes of Health) after background correction.

Solutions.

The standard extracellular Ringer's solution contained 155 mM NaCl, 4.5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM d-glucose, and 5 mM Na-Hepes (pH 7.4). Ca²⁺-free Ringer's solution with TG was prepared by substitution of 2 mM MgCl₂ and 1 mM EGTA for CaCl₂. The 20-mM Ca²⁺ Ringer's solution contained 130 mM NaCl, 4.5 mM KCl, 20 mM CaCl₂, 1 mM MgCl₂, 10 mM d-glucose, and 5 mM Na-Hepes (pH 7.4). Ca²⁺-free Ringer's solution with La³⁺ contained 155 mM NaCl, 4.5 mM KCl, 1 mM MgCl₂, 10 mM d-glucose, 10–100 μM LaCl₃, and 5 mM Na-Hepes (pH 7.4). Standard internal solution contained 150 mM Cs aspartate, 8 mM MgCl₂, 10 mM EGTA, and 10 mM Cs-Hepes (pH 7.2).

Electrophysiology.

Currents were recorded using the whole-cell patch clamp technique at 22–25 °C as described previously (24). Seals were established in Ringer's solution. After break-in, voltage stimuli (100-ms step to −100 mV, followed by 100-ms ramp from −100 to +100 mV) were delivered every 5 s from the holding potential of +30 mV. After I_CRAC reached steady state, 20 mM Ca²⁺ was introduced and peak currents were measured 3 ms after the beginning of the step to −100 mV. CDI was measured as the current decay between 3 and 195 ms after the onset of voltage steps to −60, −80, −100, and −120 mV delivered every 5 s (24). Ca²⁺-free Ringer's solution with 10–100 μM LaCl₃ was used for leak subtraction for all recordings and displayed traces.

Calcium Imaging.

Cells were loaded with 1 μM fura-2/AM (Invitrogen) for 30 min at 37 °C in DMEM. Ratiometric Ca²⁺ imaging was performed using the Axiovert 200M microscope imaging system described above. Cells were illuminated alternately at 350 and 380 nm for 40 ms every 5 s with a fura-2 filter set (400 DCLP dichroic, 480 LP emission). Illumination and image acquisition were controlled by scripts in MetaMorph (Molecular Devices). The initial rate of Ca²⁺ entry in single cells was measured 5–10 s after Ca²⁺ readdition as the slope of the F₃₅₀/F₃₈₀ fluorescence ratio and averaged in bins of 0.2 a.u. of Orai-GFP expression. All experiments were conducted at 22–25 °C.