A STIM1-dependent ‘trafficking trap’ mechanism regulates Orai1 plasma membrane residence and Ca²⁺ influx levels

Ca²⁺ influx occurs through a variety of pathways including through receptor-activated channels that open in response to different signals such as diacylglycerol (DAG), arachidonic acid and store depletion (Hofmann et al., 1999; Luo et al., 2001; Mignen et al., 2001; Parekh and Penner, 1997). Store-operated Ca²⁺ entry (SOCE) is a ubiquitous Ca²⁺ influx pathway at the cell membrane that is defined as Ca²⁺ influx in response to emptying of intracellular Ca²⁺ stores. SOCE is crucial for many physiological cell functions, including activation of cells in the immune system, secretion and muscle development (Feske, 2010; Feske et al., 2006; Liu et al., 2000; McCarl et al., 2009; Vig et al., 2008). RNA interference (RNAi) screens have identified the key molecular components of SOCE, including STIM1 and Orai1 (Liou et al., 2005; Roos et al., 2005). STIM1 is an integral single-pass endoplasmic reticulum (ER) membrane protein with lumenal EF-hand motifs that bind Ca²⁺, and as such allow STIM1 to sense store Ca²⁺ levels. Orai1 is a tetra-pass plasma membrane protein, that forms highly selective Ca²⁺ channels (Prakriya et al., 2006; Vig et al., 2006a,,b; Zhang et al., 2006). STIM1 and Orai1 are integral components of the SOCE channel, as both contribute to channel selectivity and permeation (McNally et al., 2012).

Ca²⁺ store depletion in response to receptor activation results in the clustering of STIM1 into large puncta that localize to a cortical ER domain in close proximity to the plasma membrane (∼20 nm), allowing the cytoplasmic domain of STIM1 to span that distance and physically open plasma membrane Orai1 (Liou et al., 2007; Luik et al., 2006; Park et al., 2009; Prakriya et al., 2006; Stathopulos et al., 2006; Wu et al., 2006). SOCE requires plasma membrane localization of Orai1; however, little is known about Orai1 trafficking and plasma membrane residency in mammalian cells. SPCA2 (also known as ATP2C2) has been implicated in plasma membrane localization of Orai1 in mammary epithelial cells (Cross et al., 2013). Furthermore, Orai1 trafficking has been dissected in substantial detail in Xenopus oocytes in interphase and meiosis (Yu et al., 2009,, 2010). Here, we quantitatively define Orai1 trafficking dynamics and subcellular distribution at steady state and following store depletion. We show that, at rest, ∼40% of Orai1 localizes to the plasma membrane, and that Orai1 recycles at a high rate between the intracellular and plasma membrane pools. Interestingly, store depletion is coupled to translocation of the intracellular pool of Orai1 to the plasma membrane in a STIM1-dependent fashion. At intermediate Orai1:STIM1 ratios, Orai1 translocates to and is enriched at the plasma membrane owing to a ‘trafficking trap’ mechanism that sequesters Orai1 into cortical STIM1 clusters and removes it from the plasma membrane recycling pool. In contrast, at low Orai1:STIM1 ratios (high STIM1), Orai1 does not translocate to the plasma membrane following store depletion as it becomes trapped intracellularly by the excess STIM1. Consistent with this, SOCE levels show the same biphasic dependence on the Orai1:STIM1 ratio as Orai1 plasma membrane levels. These results show that both Orai1 plasma membrane and SOCE levels are modulated based on the relative expression levels of Orai1 and STIM1. Therefore, the intracellular Orai1 pool, through plasma membrane enrichment following store depletion, modulates the rate of Ca²⁺ influx through SOCE and as such would affect downstream Ca²⁺-dependent signaling.

To study Orai1 trafficking, we used a dually tagged Orai1 with YFP at the N-terminus and an HA tag inserted in the second extracellular loop (Fig. 1B) (Park et al., 2009). This is a useful reporter because it permits the detection of plasma membrane Orai1 using anti-HA antibodies in non-permeabilized cells, whereas the YFP signal allows the normalization of the levels of surface Orai1 to total cellular Orai1. To assess the functionality of this reporter in mediating SOCE, we co-expressed YFP–HA–Orai1 with STIM1 in CHO cells, and measured SOCE by Ca²⁺ imaging (Fig. 1A). This led to significantly enhanced SOCE in response to store depletion with thapsigargin, showing that YFP–HA–Orai1 is functional (Fig. 1A). The YFP–HA–Orai1 construct has been previously used by others and shown to be also functional in other cell types (Park et al., 2009; Prakriya et al., 2006).

To quantitatively determine the subcellular distribution of Orai1 at steady state, we adapted methods previously employed for the Glut4 transporter (Karylowski et al., 2004). Absolute quantification of Orai1 plasma membrane levels requires the use of the same fluorophore in the same cells. We thus visualized Orai1 using anti-HA antibodies followed by a Cy3-conjugated secondary antibody, with plasma membrane Orai1 detected in non-permeabilized cells and total Orai1 detected in permeabilized cells (Fig. 1C). We generated stable YFP–HA–Orai1 cells out of concern that transient expression might not lead to the stabilization of the Orai1 subcellular distribution. However, whether Orai1 was expressed transiently or stably, ∼40% of the Orai1 pool localized to the plasma membrane (Fig. 1D). A similar surface and total distribution of YFP–HA–Orai1 at steady state was observed in HEK293 cells (0.397±0.040, mean±s.e.m., n=4), arguing that Orai1 trafficking in CHO and HEK cells is similar. Time-lapse imaging on live cells illustrated the dynamic localization of intracellular Orai1 with YFP–HA–Orai1-positive intracellular vesicular compartments budding off and fusing with the plasma membrane (Fig. 1E; supplementary material Movie 1).

The enrichment of Orai1 intracellularly and the dynamic Orai1 trafficking pool argue that Orai1 recycles continuously at the plasma membrane. To directly quantify the rate of Orai1 exocytosis, we incubated cells with saturating concentrations of anti-HA antibodies and measured its accumulation within cells over time – initially anti-HA antibodies will bind to plasma membrane Orai1, and at increasing times additional binding will require the translocation of intracellular Orai1 to the surface (Fig. 2A). Indeed, at longer incubation times anti-HA accumulated intracellularly (Fig. 2B). Quantifying the accumulation of cell-associated anti-HA over time can, hence, be used as a measurement of the rate of Orai1 exocytosis at steady state. Data were fitted to a mono-exponential growth function (see Materials and Methods). The Orai1 exocytosis rate constant of K_ex=0.0995±0.0173 min⁻¹ (mean±s.e.m.), indicates rapid recycling at the plasma membrane. In comparison, adipocyte Glut4 at steady state has a K_ex of ∼0.003 min⁻¹, which increases to ∼0.08 min⁻¹ only after insulin stimulation, when Glut4 translocates to the plasma membrane (Karylowski et al., 2004). Therefore, Orai1 rapidly cycles through the plasma membrane with a steady-state plasma membrane residency of ∼40%. A similar rate for YFP–HA–Orai1 exocytosis was measured in HEK293 cells (supplementary material Fig. S1).

We measured Orai1 internalization as outlined in Fig. 2D (Blot and McGraw, 2008). Cells were incubated with anti-HA antibodies for different time periods, and were then fixed and incubated with a Cy5-conjugated secondary to stain plasma membrane Orai1, followed by permeabilization and staining with a Cy3-conjugated secondary to label intracellular anti-HA-bound Orai1 that was taken up during the incubation time (Fig. 2D). At early time points in the assay, Cy5 labeling of surface Orai1 was apparent, and gradually intracellular Orai1 became Cy3 positive. The slope of the line that fitted the plotted ratio of internal (Cy3) over surface (Cy5) Orai1 provides a rate that is proportional, but not equal, to the Orai1 endocytosis rate, because two different fluorophores are used. Nevertheless, the rate of endocytosis of Orai1 was similar in transiently or stably transfected cells (Fig. 2F). Knowing the levels of surface and internal Orai1 and the exocytosis rate constant, one can calculate the endocytosis rate constant using the relationship Orai1_Surface×K_endo=Orai1_Internal×K_ex, which gives an endocytosis rate constant for Orai1 of ∼0.15 min⁻¹.

The endocytosis experiment reveals an unusual intracellular Orai1 staining pattern (Fig. 2E, Cy3-Internal) as it is targeted to intracellular vesicles that colocalize with the plasma membrane in these epifluorescence images. These vesicles were even apparent in the YFP image after anti-HA incubation (Fig. 2E). Confocal imaging provides better resolution of the distribution of Orai1 following anti-HA incubation (Fig. 3A). Surface Cy5 staining revealed a smooth Orai1 plasma membrane distribution, whereas intracellular Orai1 staining was vesicular (Fig. 3A, Cy3-Internal). A subset of these intracellular vesicles was superimposed on the plasma membrane plane at the resolution of the light microscope (Fig. 3A, close-up merge, arrow). However, the HA epitope in these vesicles had not been exposed to the extracellular space because they were not Cy5 positive, confirming their intracellular distribution (Fig. 3A). These results argue that after binding anti-HA, Orai1 accumulates in a vesicular compartment that is in very close proximity to the plasma membrane. We will refer to this pool as sub-plasmalemmal Orai1.

Because sub-plasmalemmal Orai1 becomes most prominent after anti-HA incubation, we were interested in determining whether it is also present in cells that have not been exposed to anti-HA incubation. We used line scans across both permeabilized and non-permeabilized cells stained with anti-HA antibody followed by a Cy5-conjugated secondary, which is the same approach used for the surface-to-total ratio measurement experiments (Fig. 1C,D). Line scans on non-permeabilized cells revealed enrichment of both the Cy5 and YFP signals at the plasma membrane of the cell (Fig. 3B, arrows), but practically no Cy5 labeling intracellularly, whereas YFP, as expected showed, a significant signal within the cell (Fig. 3B). Permeabilized cells exhibited a similar profile between the YFP and Cy5 signals (Fig. 3C, permeabilized). To test for the presence of sub-plasmalemmal Orai1, we measured the Cy5:YFP ratio at the plasma membrane plane, which was defined as the peak of the Cy5 signal. The plasma membrane Cy5:YFP increased significantly following permeabilization (Fig. 3D), consistent with the presence of a sub-plasmalemmal Orai1 pool, and permeabilization revealed intracellular HA–Orai1 epitopes that would be otherwise unavailable in non-permeabilized cells. In support of the ratio analysis, the width at half maximum (FWHM) of the Cy5 peak was significantly broader in permeabilized compared to non-permeabilized cells (Fig. 3E), arguing that a significantly larger portion of Orai1 close to the plasma membrane becomes available to the anti-HA antibody when cells are permeabilized. Although neither the dual staining approach (Fig. 2E), nor the linescan analyses (Fig. 3) alone is conclusive, together these results support the idea that, at steady state, a subset of the Orai1 pool localizes to a sub-plasmalemmal membrane vesicular compartment.

The large intracellular Orai1 pool present at rest could act as a reservoir to increase Orai1 plasma membrane levels and, as such, SOCE, following store depletion. Indeed store depletion led to a significant enrichment of YFP–HA–Orai1 as measured using the surface-to-total ratio assay (Fig. 4A). As these are steady-state measurements, they show that store depletion results in the translocation to, and stabilization of, intracellular Orai1 at the plasma membrane. We postulated that Orai1 plasma membrane enrichment following store depletion is due to trapping of recycling Orai1 by activated STIM1 clusters in the cortical ER (Fig. 4B). Store depletion would lead to the co-clustering of resident plasma membrane Orai1 with cortical STIM1 as additional Orai1 molecules translocate from the intracellular compartment to the plasma membrane and become trapped into STIM1 puncta and as such would be no longer available to the internalization machinery (Fig. 4B), a model we will refer to as the ‘trafficking trap’ model.

We tested Orai1 enrichment following store depletion in stable YFP–HA–Orai1 cells. For the proposed trafficking trap model to work under these conditions, STIM1 cannot be limiting. The role of STIM1 expression levels is analyzed in substantial detail below. However, western blots of endogenous STIM1 and Orai1 are consistent with the idea that STIM1 is not limiting in CHO cells (Fig. 4C).

To further assess our model, we tested whether endogenous Orai1 was also enriched at the plasma membrane following store depletion by assessing biotinylation of plasma membrane proteins. Store depletion (mediated by thapsigargin) was associated with increased levels of surface Orai1 (Fig. 4D), showing that endogenous Orai1 is enriched at the plasma membrane in a similar fashion to overexpressed YFP–HA–Orai1. The increase in plasma membrane Orai1 was specific, as no change in the PMCA (also known as ATP2B) levels were observed in thapsigargin-treated cells (Fig. 4D). The specificity of the biotinylation approach to measure plasma membrane proteins is illustrated by the lack of actin reactivity in the biotinylated fraction. Furthermore, in the absence of biotin, no reactivity against Orai1 or PMCA was observed (Fig. 4D), showing that biotinylated plasma membrane proteins were pulled down. Similar findings have been previously reported in platelets, where store depletion is associated with Orai1 plasma membrane enrichment (Woodard et al., 2008).

Another tenant of trafficking trap model is that, following store depletion, Orai1 that is trapped in the STIM1 puncta would be no longer engaged in continuous cycling at the plasma membrane. To directly test this postulate, we measured the rate of Orai1 exocytosis in control and thapsigargin-treated cells. To avoid any contribution from the anti-HA-antibody-binding kinetics, we depleted Ca²⁺ stores with thapsigargin and then incubated cells with the anti-HA antibody at 4°C to allow for anti-HA binding while trafficking is blocked. Cells were then released to 37°C and cell-associated anti-HA measured (Fig. 4E). With Ca²⁺ stores full, Orai1 showed the typical exocytotic cycling (Fig. 4E, Con). In contrast, when Ca²⁺ stores were depleted, the majority of the Orai1 pool localized to the plasma membrane and no longer cycled (Fig. 4E, thapsigargin). These data show that following store depletion Orai1 does not cycle and is trapped at the plasma membrane, presumably through its association with STIM1. Hence, cortical STIM1 puncta formed in response to store depletion act as a sink that traps Orai1 molecules, thus removing them from the plasma-membrane–intracellular cycling loop and enriching Orai1 at the plasma membrane. The Orai1 K_ex rate constant measured using the 4°C incubation followed by release to 37°C was 0.169±0.011 min⁻¹ (mean±s.e.m.) (Fig. 4E), which is higher than that measured following direct incubation of anti-HA at 37°C (Fig. 2C). This argues that anti-HA binding kinetics contribute to an underestimation of the Orai1 exocytosis rate.

Store depletion with thapsigargin is not physiological, leading to the possibility that the observed Orai1 plasma membrane enrichment might be an artifact of the irreversible store depletion induced by thapsigargin. To test Orai1 trafficking in response to a physiological stimulus, HEK cells were treated with the cholinergic agonist carbachol (CCh) to mobilize Ca²⁺ and activate SOCE (Fig. 4F) (DeHaven et al., 2009). CCh or methacholine (MCh) treatment results in Orai1 plasma membrane enrichment to levels comparable to those observed with thapsigargin (Fig. 4G), showing that Orai1 plasma membrane translocation in response to store depletion is physiological.

The Orai1 trafficking trap model predicts a dependency on STIM1 to enrich Orai1 at the plasma membrane following store depletion. If STIM1 became limiting it would be expected to abrogate its ability to trap recycling Orai1 at the plasma membrane. To test this premise, we knocked down endogenous STIM1 using small interfering RNA (siRNA), leading to a strong (>80%) reduction in protein levels (Fig. 5A). As predicted by the model, limiting STIM1 levels prevented plasma membrane enrichment of Orai1 following store depletion (Fig. 5B). In contrast, cells treated with control siRNA exhibited the typical 1.3–1.4-fold enrichment of plasma membrane Orai1 (Fig. 5B), showing that the siRNA treatment in itself does not negatively impact upon Orai1 plasma membrane enrichment.

To further test the model, we overexpressed STIM1 (1 µg DNA transfection) in cells stably expressing Orai1. Surprisingly, cells overexpressing STIM1 did not exhibit any plasma membrane enrichment of Orai1 following store depletion, but rather showed a tendency for decreased levels of plasma membrane Orai1 enrichment (Fig. 5C). This contradicts the trafficking trap model, which would argue that the presence of excess STIM1 should result in a more pronounced plasma membrane enrichment of Orai1 given the expected increase in STIM1 puncta formation. Therefore, Orai1 plasma membrane enrichment in response to store depletion shows a biphasic dependence on STIM1 expression levels with no enrichment at low (siRNA) or high (1 µg DNA transfection) STIM1 expression levels (Fig. 5D). For Fig. 5D, estimates of the Orai1:STIM1 ratio under the three conditions (1 µg STIM1 expression, control stable Orai1-expressing cells and cells expressing STIM1 siRNA) were obtained from the western blotting data and quantification shown in Fig. 5I,J, as discussed below.

To further explore the biphasic dependence of Orai1 plasma membrane enrichment on STIM1 expression levels, we wanted to develop an assay that correlated the Orai1:STIM1 ratio to Orai1 plasma membrane levels over a broader range of STIM1 and Orai1 expression than is possible with the surface-to-total ratio assay. We therefore resorted to flow cytometry, which allows rapid analysis of thousands of cells over at least three orders of magnitude of expression levels. Plasma membrane Orai1 was detected by using anti-HA antibodies followed by APC-labeled secondary. Consistent with the surface-to-total ratio assay results, in stable YFP–HA–Orai1 cells without STIM1 transfection, the Orai1 population showed an increased APC fluorescence indicative of higher plasma membrane Orai1 levels following store depletion (Fig. 5E; supplementary material Fig. S2). Interestingly, when cells were transfected with 0.2 µg STIM1 DNA, Orai1 plasma membrane enrichment following store depletion was significantly more pronounced (Fig. 5F; supplementary material Fig. S2). This is consistent with the trafficking trap model where higher STIM1 levels are associated with increased plasma membrane Orai1 after SOCE activation. Also consistent with the model is the finding that STIM1 siRNA abrogated plasma membrane Orai1 enrichment after store depletion (Fig. 5H, supplementary material Fig. S2). In contrast, and consistent with the results obtained using the surface-to-total ratio assay, when cells were transfected with 1 µg STIM1 plasmid, store depletion was associated with a decrease in Orai1 plasma membrane levels rather than an increase (Fig. 5G; supplementary material Fig. S2).

The dependence on STIM1 for the Orai1 translocation to the plasma membrane in response to store depletion was also apparent within a single flow cytometry experiment using cells transfected with 1 µg STIM1 plasmid (Fig. 5K). The surface-to-total ratio of Orai1 (APC–HA–Orai1:YFP–Orai1) was plotted as a function of the Orai1:STIM1 ratio, either as individual cells (Fig. 5K, left panel) or as the average of bins at 0.5 Orai1:STIM1 ratio increments (Fig. 5K, right panel). This analysis shows that cells with high STIM1 expression (Orai1:STIM1 ratio <1) exhibited low levels of plasma membrane Orai1 following store depletion as compared to cells with lower STIM1 expression within the same population (Fig. 5K).

To estimate the levels of Orai1 and STIM1 under the various conditions outlined above (i.e. knockdown of endogenous STIM1 or overexpression of STIM1 and/or Orai1), we resorted to western blot analysis using the quantitative Licor platform (Fig. 5I,J). Lysates from cells under the various conditions were separated and both endogenous and overexpressed STIM1 and Orai1 were detected (Fig. 5I). Quantification of these expression levels from three different experiments is shown in Fig. 5J. The same data normalized to the control condition are illustrated in supplementary material Fig. S3. Plotting the Orai1 surface-to-total ratio as a function of the Orai1:STIM1 ratio as quantified from the western blot (Fig. 5J), shows that the ability of Orai1 to be enriched at the plasma membrane following store depletion is dependent on the Orai1:STIM1 ratio in a biphasic fashion, with minimal Orai1 enrichment at the extremes of STIM1 expression (Fig. 5L).

To better understand the biphasic dependence of Orai1 plasma membrane enrichment on STIM1 levels, we imaged STIM1 and Orai1 subcellular distribution in cells stably expressing YFP–HA–Orai1 and transfected with mCherry–STIM1 (0.2 or 1 µg STIM1 plasmid) by confocal microscopy (Fig. 6A, Con STIM1). At focal planes deep within the cell, the YFP and anti-HA signals colocalized (Fig. 6A, Con STIM1). Store depletion in cells transfected with 0.2 µg STIM1 led to the formation of coincident STIM1 and Orai1 clusters, with the Orai1 clusters localizing at the plasma membrane (they were colabeled with the anti-HA antibody, which only stains plasma membrane Orai1) (Fig. 6A). Store depletion in cells transfected with 1 µg STIM1 also led to the formation of coincident STIM1–Orai1 puncta, but interestingly, in this case, the majority of the STIM1–Orai1 clusters localized intracellularly given that they were not labeled with the anti-HA antibody (plasma membrane Orai1) (Fig. 6A). This is apparent in the merged image of the YFP and Cy5–HA signals (Fig. 6A, arrows). This argues that STIM1 traps Orai1 intracellularly and prevents it from reaching the plasma membrane when it is expressed in significant excess of Orai1, as in cells transfected with 1 µg STIM1 (Fig. 5I,J). These observations are supported by statistical analysis of the colocalization of plasma membrane Orai1 with total Orai1 using both the Pearson and Manders coefficients. For this analysis, we quantified plasma membrane and total Orai colocalization both at the plasma membrane focal plane right above the coverslip, and at a focal plane at the center of the cell (deep focal planes) after Ca²⁺ store depletion. This was performed in cells transfected with 0.2 µg or 1 µg STIM1 plasmid DNA. Pearson's analyses show high colocalization at both the plasma membrane and deep focal planes in cells transfected with 0.2 µg STIM1 plasmid, consistent with the translocation and enrichment of Orai1 at the plasma membrane following store depletion (Fig. 6B). In contrast, lower levels of colocalization was detected in cells transfected with 1 µg STIM1 plasmid, and this was more pronounced at deep focal planes (Fig. 6B). This is consistent with the intracellular trapping of Orai1–STIM1 clusters.

Manders analyses allow measurement of the overlap coefficient of two fluorophores from confocal images (Manders et al., 1993). As expected the percentage of plasma membrane Orai1 (HA) colocalizing with total Orai1 (YFP) is high regardless of the levels of transfected STIM1 or focal plane (Fig. 6C). However, when the overlap coefficient of total Orai1 (YFP) with plasma membrane Orai1 (HA) was measured, high colocalization was observed at the plasma membrane and deep focal planes in cells transfected with 0.2 or 1 µg STIM1 plasmid (Fig. 6D). However, in the case of 1 µg STIM1 transfection, lower colocalization was apparent at the plasma membrane plane and, more importantly, little to no colocalization could be detected in deep planes where intracellular Orai1 would be more prominent (Fig. 6D). These data show that store depletion in cells expressing excessive levels of STIM1 (1 µg STIM1 transfection; low Orai1:STIM1) leads to the formation of intracellular STIM1–Orai1 clusters, and as such to the trapping of Orai1 intracellularly, thus preventing its translocation and enrichment at the plasma membrane. Importantly, no changes in the Manders colocalization coefficient for total-to-surface Orai1 was observed in control cells transfected with 1 µg STIM1 plasmid before store depletion (supplementary material Fig. S4). This shows that intracellular trapping of Orai1 in cells expressing excessive levels of STIM1 requires store depletion and ‘activation’ of STIM1 to bind to and trap Orai1 intracellularly.

To assess whether Orai1 plasma membrane levels following store depletion correlate with SOCE magnitude following store depletion, we measured the rate of Ca²⁺ influx as a function of the Orai1:STIM1 ratio in individual cells stably expressing Orai1 and transfected with STIM1 (Fig. 6E). Ca²⁺ stores were depleted with thapsigargin in Ca²⁺-free medium followed by Ca²⁺ addition to measure SOCE (as in Fig. 4F). The rate of Ca²⁺ influx was measured as the slope of the rising Ca²⁺ influx signal at 10–15 s after Ca²⁺ addition (Hoover and Lewis, 2011). The Ca²⁺ influx rates showed a biphasic dependence on the Orai1:STIM1 ratio (Fig. 6E), in a similar fashion to Orai1 plasma membrane enrichment. At a low Orai1:STIM1 ratio (high STIM1) the rate of Ca²⁺ influx was low; this increased at intermediate STIM1 levels to decrease again when STIM1 became limiting (high Orai1:STIM1) (Fig. 6E). These functional data (Fig. 6E) mirror the dependence of Orai1 plasma membrane enrichment on STIM1 expression levels (Fig. 5D,J), and importantly show that the translocation of intracellular Orai1 to the plasma membrane following store depletion is functionally coupled to increased levels of Ca²⁺ influx through SOCE.

Specificity in Ca²⁺ signaling is encoded in the spatial, temporal and amplitude features of Ca²⁺ signals, which are then decoded by Ca²⁺-dependent effectors into cellular function. SOCE is a ubiquitous Ca²⁺ influx pathway that is crucial for various physiological processes. Agonist-dependent Ca²⁺ signals are often linked to SOCE downstream of Ca²⁺ release from stores. SOCE contributes to shaping Ca²⁺ dynamics, leading to specific activation of downstream effectors (Courjaret and Machaca, 2014; Dolmetsch et al., 1997; Kar et al., 2012,, 2014; Parekh, 2009). SOCE levels depend on the stoichiometry of STIM1 and Orai1 in plasma membrane clusters (Hoover and Lewis, 2011), and as such on the levels of Orai1 at the plasma membrane. However, little is known about Orai1 trafficking at the plasma membrane. Here, we quantitatively define the dynamics of Orai1 trafficking both at steady state and following store depletion. Orai1 recycles rapidly at the plasma membrane with ∼40% of the total Orai1 pool residing at the plasma membrane at steady state (Fig. 1). The exocytosis rate of Orai1 (K_ex ∼0.1 min⁻¹) is high compared to other recycling proteins, including the Glut4 transporter at rest (10–30-fold higher) and after insulin stimulation (7–25-fold higher) in adipocytes and muscle (Karlsson et al., 2009; Karylowski et al., 2004). Furthermore, intracellular Orai1 localizes to a new vesicular compartment that is in close proximity to the plasma membrane (Fig. 3), which might explain its rapid recycling. The nature of this intracellular Orai1-positive vesicular compartment is unknown and will require further studies to define.

Interestingly, the Orai1 plasma membrane residence is modulated by Ca²⁺ store content in a STIM1-dependent fashion. Store depletion is coupled to a significant enrichment of both overexpressed and endogenous Orai1 at the plasma membrane (Fig. 4). We show that this enrichment is due to a trafficking trap mechanism (Fig. 7). Store depletion leads to the formation of STIM1 clusters in the cortical ER, which recruit and open plasma membrane Orai1. The trafficking trap model postulates that the plasma membrane Orai1 pool trapped within STIM1 clusters is no longer available to the endocytic trafficking machinery and, as such, is taken away from the recycling pool, leading to enrichment of plasma membrane Orai1 (Fig. 7, intermediate STIM1). In support of this model, direct measurement of Orai1 recycling shows that the plasma-membrane-enriched Orai1 no longer recycles in response to store depletion (Fig. 4E). Importantly, Orai1 plasma membrane enrichment is also observed following physiological activation of SOCE downstream of cholinergic receptor stimulation (Fig. 4). Therefore, continuous recycling of Orai1 at the plasma membrane at rest (when Ca²⁺ stores are full) results in a steady-state distribution with 40% of the total Orai1 pool residing at the plasma membrane. Store depletion results in the formation of STIM1 clusters at ER–plasma-membrane junctions that trap plasma membrane Orai1 and remove it from the recycling machinery. Hence, under conditions where STIM1 expression is in excess of Orai1, store depletion is associated with Orai1 plasma membrane enrichment owing to the trapping of Orai1 within cortical ER STIM1 clusters. Importantly, this model does not invoke any alterations in the dynamics of Orai1 trafficking associated with its enrichment at the plasma membrane following store depletion.

When STIM1 expression is knocked down using siRNA, store depletion is no longer coupled to Orai1 plasma membrane translocation. Consistent with the trafficking trap model, expression of STIM1 (0.2 µg) in stable YFP–HA–Orai1 cells, which express high Orai1 levels, leads to a more pronounced enrichment of Orai1 at the plasma membrane following store depletion (Fig. 5E,L). However, when STIM1 is expressed at high levels (1 µg STIM1 plasmid transfection), Orai1 translocation to the plasma membrane following store depletion is abrogated (Fig. 5C,D,H,K,L) owing to the formation of intracellular STIM1–Orai1 clusters that are trapped intracellularly, thus preventing Orai1 plasma membrane translocation (Fig. 6). Hence, under these conditions of excess STIM1 compared to Orai1 the trafficking trap mechanism is at play but instead of trapping Orai1 at the plasma membrane, Orai1 is instead trapped intracellularly through its association with intracellular activated STIM1 clusters.

Therefore, the levels of Orai1 protein at the plasma membrane following store depletion depend on STIM1 expression levels. Assuming constant levels of Orai1, at low levels of STIM1 expression, STIM1 is limiting and is unable to recruit intracellular Orai1 to the plasma membrane (Fig. 7, low STIM1). At intermediate STIM1 levels, the trafficking trap model explains Orai1 enrichment at the plasma membrane following store depletion (Fig. 7, intermediate STIM1). When cells express excess STIM1 compared to Orai1, STIM1 clusters are stabilized in an ER compartment deep within the cytosol that is distinct from the cortical ER, and trap intracellular Orai1, thus preventing it from reaching the plasma membrane. This ER compartment could very well represent the ‘precortical ER’ as elegantly defined by Orci et al. using cryo-electron microscopy as being enriched in STIM1 and lacking any contact with the plasma membrane (Orci et al., 2009). One possibility is that high STIM1 expression levels lead to saturation of the cortical ER compartment and the localization of STIM1 clusters to the pre-cortical ER. However, this does not seem to be the case because overexpression of STIM1 is associated with increased cortical ER formation (Orci et al., 2009). Rather it seems that at high levels of STIM1 expression, STIM1 in the pre-cortical ER, which is activated by store depletion, binds intracellular vesicular Orai1 leading to the trapping of the complex intracellularly and preventing STIM1 translocation to the cortical ER and Orai1 trafficking to the plasma membrane.

To assess whether Orai1 plasma membrane residency modulates SOCE levels after store depletion, we measured the rate of Ca²⁺ influx as a function of the Orai1:STIM1 ratio (Fig. 6E). Ca²⁺ influx mirrors the dependence on Orai1:STIM1 ratio observed for Orai1 plasma membrane enrichment (Fig. 6E). Ca²⁺ influx responds to the Orai1:STIM1 expression ratio in a biphasic fashion with low Ca²⁺ influx rates at the extremes and high rates at intermediate Orai1:STIM1 (Fig. 6E). Similar results were obtained by Kilch et al. (Kilch et al., 2013) and by Hoover and Lewis, who elegantly correlated the stoichiometry of STIM1:Orai1 within STIM1–Orai1 puncta at the plasma membrane with SOCE levels (Hoover and Lewis, 2011). They showed that Orai1 trapping within STIM1 puncta at the plasma membrane requires one or two STIM1 molecules to bind to the presumed Orai1 tetramer (a Orai1:STIM1 stoichiometry of 4:2); however, this trapping is not sufficient to open Orai1 as maximal SOCE activity is achieved at an Orai1:STIM1 stoichiometry of 1:2 within puncta at the plasma membrane (Hoover and Lewis, 2011). However, these authors did not consider the contribution of Orai1 trafficking to SOCE levels. Here, we show that Orai1 plasma membrane residence and SOCE levels depend on STIM1 expression levels in a biphasic manner. Maximal Orai1 plasma membrane enrichment and SOCE are observed at an intermediate Orai1:STIM1 ratio, and decrease at a low and high Orai1:STIM1 ratio. Therefore, the intracellular Orai1 pool provides a reservoir that modulates Orai1 plasma membrane residence following store depletion and modulates SOCE based on the Orai1:STIM1 expression ratio. Modulation of the rate of Ca²⁺ influx based on Orai1 trapping at the plasma membrane would regulate SOCE levels and the downstream Ca²⁺-dependent processes that regulate cell function.

Thapsigargin, Dulbecco's modified Eagle's medium with nutrient mixture F-12 (DMEM/F12), Fura 2-AM, Cy3- and Cy5-conjugated secondary antibodies, and penicillin-streptomycin were from Invitrogen (Carlsbad, CA). Methacholine, fetal bovine serum (FBS), rabbit polyclonal anti-Orai1 antibody, mouse anti-actin antibodies and protease inhibitor cocktails were from Sigma (St Louis, MO). Rabbit polyclonal anti-STIM1 antibody was from Cell Signaling Technology, Inc. (Danvers, MA). Mouse anti-HA antibody was obtained from Covance (San Diego, CA). Mouse monoclonal anti-PMCA antibody was from Santa Cruz Biotechnology (Santa Cruz, CA). Carbachol was obtained from Calbiochem (San Diego, CA). Horseradish-peroxidase-conjugated goat anti-rabbit-IgG and goat anti-mouse-IgG antibodies were from Jackson ImmunoResearch Laboratories (West Grove, PA). Enhanced chemiluminescence (ECL) detection reagents, immobilized streptavidin gel and EZ-Link Sulpho-NHS-LC-Biotin were from Pierce (Rockford, IL). IRDye^®800CW goat anti-mouse-IgG and IRDye^®680RD goat anti-rabbit-IgG antibodies were obtained from LI-COR Biosciences (Lincoln, NE). Clone pDS-YFP-HA-Orai1 was a kind gift from Rich Lewis (Stanford University, Palo Alto, CA) (Park et al., 2009; Prakriya et al., 2006).

Protein extracts were obtained from cells lysed in cold lysis buffer containing 25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% Triton X-100 and protease inhibitors for 20 min, followed by centrifugation at 9,400 g for 10 min at 4°C. Protein concentrations were determined using a Bio-Rad protein assay (Hercules, CA). Equal amounts of proteins were loaded on NuPAGE^® Novex^® 4-12% Bis-Tris Protein Gels (Invitrogen). Electrophoresed samples were transferred onto PVDF membrane and blocked with 5% milk in TBST, and the membrane was incubated overnight at 4°C with primary antibody with constant shaking. After washing, horseradish-peroxidase-conjugated secondary antibody was applied. Proteins bound to the secondary antibody were visualized using ECL (Amersham Biosciences, Piscataway, NJ). Two-color western blotting was performed using the secondary IRDye^®800CW goat anti-mouse-IgG and IRDye^®680RD goat anti-rabbit-IgG antibodies, and blots were imaged on a LI-COR Odyssey platform followed by quantification using LI-COR Image Studio Lite v. 4.0.

CHO cells were obtained from the ATCC (CCL-61) and human embryonic kidney (HEK293) cells were cultured at 37°C and 5% CO₂ in high-glucose DMEM/F12 and high-glucose DMEM (Invitrogen), respectively. Media for both cell lines were supplemented with 10% heat-inactivated FBS (Invitrogen), 5% (v/v) of a stock solution containing 5000 IU ml⁻¹ penicillin and 5000 μg ml⁻¹ streptomycin (Invitrogen). In some studies, the TRVb1 CHO cell line was used, which lacks the endogenous transferrin receptor and stably expresses the human transferrin receptor (McGraw et al., 1987). CHO and HEK293 cells, grown to 50–70% confluence, were transiently transfected with plasmid pDS-YFP-HA-Orai1 DNA, containing the human Orai1 (GenBank number NM_032790). Transient transfections were performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Stable expression cell lines were generated by pDS-YFP-HA-Orai1 DNA transfection and selected using selective medium (DMEM, 10% FBS, 800 µg/ml G418, 100 µg/ml penicillin and streptomycin). Stable clones were sorted by flow cytometry and checked for protein expression using fluorescence microscopy and western blotting.

YFP–HA–Orai1 stable cells were transfected with the siRNA ON-Target plus Smart pool (50 nM) designed against mouse STIM1 (NM_009287), or non-targeting siRNA as a control, using DharmaFECT transfection reagent (Dharmacon). Four siRNA duplexes targeting mouse STIM1 were used: 5′-GAAGUAGGCAGACUAGGGU-3′, 5′-AAACAUAGCACCUUCCAUG-3′, 5′-GAUCGGAGCCACAGGCAGA-3′, and 5′-UACAGUGGCUCAUUACGUA-3′. These duplexes have 78–100% sequence identity to hamster STIM1 and effectively knockdown STIM1 in CHO cells (Fig. 5A). Cells were incubated for 16 h with the respective siRNA; the transfection medium was aspirated and replaced with fresh medium without siRNA for 48 h before imaging or western blot analysis.

Live-cell imaging was performed on a confocal microscope (Leica) using a 63×1.4 NA oil objective at room temperature. Cells were grown on poly-D-lysine-coated glass-bottomed plates (MatTek Corporation). YFP, Cy3 and Cy5 were excited with the 514, 561 and 633 nm laser lines, respectively. In all imaging experiments, gain and offset settings were adjusted for each experiment using non-transfected control cells. For kinetic measurements, fluorescent images were collected on a DMIRB inverted microscope (Leica Microsystems, Deerfield, IL) using a 20× objective. Fluorescence quantifications were performed with MetaMorph image processing software (Molecular Devices, Sunnyvale, CA). Time-lapse imaging was performed by acquiring images every 5 s and was processed using ImageJ (National Institutes of Health).

For imaging Orai1 and STIM1 subcellular localization, YFP–HA–Orai1 stable CHO cells plated on poly-D-lysine coated glass-bottomed plates (MatTek Corporation) were transfected with 0.2 µg pCMV-XL6-mCherry-STIM1 DNA, or 1 µg pCMV-XL6-mCherry-STIM1 DNA using Lipofectamine 2000. After 48 h, cells were washed with Ca²⁺-free PBS before incubating in Ca²⁺-free PBS containing 1 µM thapsigargin for 20 min at 37°C. Cells were then fixed with 4% paraformaldehyde (PFA), washed with Ca²⁺-free PBS, and incubated with mouse anti-HA antibody for 45 min at 37°C. Following primary antibody incubation, cells were washed and stained with Cy5-conjugated anti-mouse-IgG antibody. Imaging was performed on a confocal microscope (Leica) using a 63×1.4 NA oil lens. Protein colocalization was assessed in 15–17 cells for each treatment by determination of the Pearson's and Manders coefficients using ImageJ software (http://rsbweb.nih.gov/ij) on images captured at the focal plane just above coverslip (plasma membrane) and at the focal plane at the center of the cell (deep focal plane).

Cells were fixed using freshly prepared 4% PFA (w/v) for 10 min at room temperature with our without permeabilization for 10 min in PBS containing 0.1% (w/v) Triton X-100. All subsequent antibody incubations were performed in PBS containing 5% (w/v) FBS. Cells were stained with saturating concentrations of monoclonal anti-HA antibodies (Covance) at 1:200 for 45 min at 37°C, followed by incubation in a 1:400 dilution of Cy3-conjugated goat anti-mouse-IgG (Invitrogen) for 30 min at 37°C. The ratio Cy3:YFP was determined by quantitative fluorescence microscopy as follows. Fields with cells expressing the HA–Orai1–YFP were randomly chosen in the YFP channel blinded to the expression of HA–Orai1–YFP on the plasma membrane (Cy3 channel). Images were collected in both YFP and Cy3 channels. Prior to collecting the data, exposure times for the channels were independently set to maximize the signal while minimizing the number of cells with expression levels above saturation for detection, thereby optimizing the dynamic range of the assay. Once set for each channel, all images in that channel were collected at the same exposure. Metamorph or Image J software was used to quantify fluorescence of YFP and Cy3 at the single-cell level. To correct for background, cells that did not express the HA–Orai1–YFP construct, identified from the YFP channel, were circled and fluorescence in the channels was also logged. Typically, two background cells per field were collected. The background corrections were derived from cells in the same dishes as the experimental cells, thereby controlling for background due to non-specific binding of the primary and secondary antibodies. The average background fluorescence per pixel in each channel was calculated for each dish and this value was subtracted from the fluorescence of the expressing cells in that dish. The ratio of the background-subtracted Cy3 fluorescence per pixel divided by the YFP fluorescence per pixel was determined, yielding a surface-to-total ratio per cell. Those values were averaged to yield a surface-to-total HA–Orai1–YFP value per condition. For studies of transiently transfected cells, non-transfected cells (no YFP expression) were used for background calculation. For studies of stable expressing cells, non-transfected cells were spiked into the cultures at the time of plating to be used for background determination. The fraction of plasma membrane Orai1 was determined by dividing the Cy3/YFP for non-permeabilized cells by the Cy3/YFP value for permeabilized cells.

where A is the plateau of maximal uptake; B the amplitude of the signal; t the time and K_ex the exocytosis rate constant.

To measure the amount of Orai1 internalized as a function of time, cells were incubated with saturating concentrations of anti-HA antibody and, at various times, cells were fixed and stained with Cy5-conjugated secondary to label plasma membrane Orai1. Cells were fixed again, permeabilized and stained with Cy3-conjugated secondary to reveal HA-bound Orai1 that had been internalized. The same anti-mouse-IgG antibody was used with a Cy5 or Cy3 label so that Cy5-conjugated anti-mouse-IgG blocks all epitopes on the plasma membrane, thereby limiting Cy3 labeling to internal HA-labeled Orai1. Cy3 and Cy5 signals were measured by quantitative fluorescence microscopy. The ratio (Cy3/YFP)/(Cy5/YFP) was plotted over time and fitted with a linear regression.

Control or thapsigargin-treated cells (1 μM, 15 min; to deplete intracellular Ca²⁺ stores) were washed with ice-cold Söerscen's buffer containing 16 mM Na₂HPO₄ and 114 mM NaH₂PO₄, pH 8.0. Cell surface proteins were labeled by re-suspending in EZ-Link Sulpho-NHS-LC-Biotin (2.5 mg in 12 ml ice-cold Söerscen's buffer) and incubated while mixing for 1 h at 4°C. The biotinylation reaction was terminated by addition of 100 μl of 1 M Tris base, and the remaining biotinylating agent was removed by washing the cells in ice-cold Söerscen's buffer. Cells were subsequently lysed with RIPA buffer (25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and protease inhibitors) and labeled proteins were isolated with agarose beads coated with streptavidin, a biotin-binding protein. The bound proteins were released by incubating with SDS-PAGE sample buffer and were subjected to western blotting.

For measuring surface Orai1 levels, YFP–HA–Orai1 stable CHO cells were transiently transfected with 0.2 or 1 µg pCMV-XL6-mCherry-STIM1 plasmid DNA, or with 50 nM STIM1 siRNA or non-targeting siRNA, and cultured for 48 h after transfection. Cells were washed with Ca²⁺-free PBS prior to addition of Ca²⁺-free PBS containing 1 µM thapsigargin (Invitrogen) for 15 min at 37°C; then PBS was removed and cells were quickly fixed by addition of 4% PFA for 10 min at room temperature. Cells were then washed with Ca²⁺-free PBS, scraped gently from plates, and pelleted at 400 rpm for 10 min. Cells were then incubated with anti-HA antibody (Covance) for 45 min at 37°C. Following washing with Ca²⁺-free PBS, cells were stained with APC-conjugated secondary antibody (Invitrogen) for 30 min at 37°C. Cells were then analyzed for GFP, mCherry and APC by flow cytometry (BD FACSAria). Data from ∼10,000 GFP-positive cells per sample were analyzed either using the BD software or were extracted using FlowPy, and APC:GFP and mCherry:GFP ratios were analyzed using Origin 8.5. The experiment was repeated at least three times.

All the quantitative values are expressed as mean±s.e.m. Statistical significance was performed using analysis of variance and Student's t-test or ANOVA. Differences are considered significant at P<0.05 and are indicated in the figures as follows: *P<0.05, **P<0.01 and ***P<0.001.

We thank the Flow Cytometry Facility within the Microscopy Core at Weill Cornell Medical College in Qatar for contributing to these studies. The Core is supported by the ‘Biomedical Research Program at Weill Cornell Medical College in Qatar’, a program funded by Qatar Foundation. The statements made herein are solely the responsibility of the authors.