Sigma1 receptors inhibit store-operated Ca²⁺ entry by attenuating coupling of STIM1 to Orai1

Sigma1 receptors (σ1Rs) are widely distributed in the brain and peripheral tissues, including lung, kidney, liver, and spleen, and highly expressed in some tumor cells (Walker et al., 1990; Vilner et al., 1995; Monnet, 2005; Stone et al., 2006; Cobos et al., 2008; Wu and Bowen, 2008; Su et al., 2010; Brune et al., 2013). They are regulated by an unusually diverse array of ligands, including endogenous steroids, drugs of abuse such as cocaine and methamphetamine, and drugs used to treat depression, anxiety, psychosis, pain, and neurodegenerative diseases (Maurice et al., 1999; Waterhouse et al., 2007; Maurice and Su, 2009; Su et al., 2010; Navarro et al., 2012; Robson et al., 2012; Wünsch, 2012; Kourrich et al., 2013; Tsai et al., 2014). Changes in expression and polymorphisms of σ1Rs are associated with heart failure (Ito et al., 2012, 2013), addiction (Maurice et al., 2002; Kourrich et al., 2013), neurodegenerative and psychiatric disorders (Miki et al., 2014; Tsai et al., 2014), and cancer (Spruce et al., 2004; Wang et al., 2004; Aydar et al., 2006; Maurice and Su, 2009; Crottès et al., 2013). These associations have provoked interest in σ1Rs as both therapeutic targets and diagnostic tools (van Waarde et al., 2015).

The σ1R is an integral membrane protein with two transmembrane domains. It is expressed in the ER, where it is concentrated in cholesterol-rich mitochondrion-associated ER membrane (MAM) domains and bound to the ER luminal chaperone, BiP (Fig. S1; Hayashi and Su, 2003, 2007; Palmer et al., 2007). Agonists of σ1R cause it to dissociate from BiP and MAM, allowing σ1Rs to move within ER membranes and interact with signaling proteins in the plasma membrane (PM), most notably ion channels, thereby regulating their activity (Su et al., 2010; Balasuriya et al., 2012; Pabba, 2013). Antagonists block this effect (Fig. S1 and Table S1). Loss of Ca²⁺ from the ER can also release σ1Rs from their interaction with BiP, freeing them to interact with other proteins (Hayashi and Su, 2007). In addition to regulating the activity of these proteins, σ1Rs can also act as chaperones, stabilizing signaling proteins as they traffic along the secretory pathway (Tsai et al., 2014). σ1Rs may also be expressed in the nuclear envelope (Hayashi and Su, 2005a,b; Brune et al., 2013; Mori et al., 2013) and PM (Lupardus et al., 2000; Aydar et al., 2002; Brune et al., 2013; Balasuriya et al., 2014a) and may even be secreted into the extracellular space (Hayashi and Su, 2003; Su et al., 2010). The interactions between σ1Rs and ion channels may therefore occur within the plane of a membrane (ER or PM) or across ER–PM junctions (Hayashi and Su, 2007; Kourrich et al., 2013; Balasuriya et al., 2014a). Clearly, σ1Rs are important links between diverse ligands, physiological stimuli, and many key signaling molecules (Hayashi and Su, 2007; Su et al., 2010; Kourrich et al., 2013).

Receptors that stimulate PLC and formation of inositol 1,4,5-trisphosphate (IP₃) evoke both Ca²⁺ release from the ER through IP₃ receptors (IP₃Rs) and Ca²⁺ entry across the PM. At MAMs, Ca²⁺ released by IP₃Rs can be rapidly accumulated by mitochondria, thereby stimulating oxidative phosphorylation (Rizzuto et al., 2012) and promoting cell survival (Cárdenas et al., 2010), whereas excessive mitochondrial Ca²⁺ uptake triggers apoptosis (Mallilankaraman et al., 2012). The association of IP₃R3s with σ1Rs at MAMs supports the transfer of Ca²⁺ from the ER to mitochondria by curtailing the degradation of active IP₃R3s (Hayashi and Su, 2007). The increase in mitochondrial Ca²⁺ concentration and resultant boost in oxidative phosphorylation are thought to contribute to the prosurvival effects of σ1Rs in the central nervous system and cancer cells (Lewis et al., 2014). One effect of σ1Rs may therefore be to support transfer of Ca²⁺ from the ER to mitochondria, but this transfer also depends on the Ca²⁺ content of the ER.

The Ca²⁺ entry evoked by receptors that stimulate PLC is usually mediated by store-operated Ca²⁺ entry (SOCE), which is stimulated by loss of Ca²⁺ from the ER (Parekh and Putney, 2005; Hogan and Rao, 2015). The reduction in Ca²⁺ concentration within the ER is detected by the luminal EF hands of stromal interaction molecule 1 (STIM1), an integral ER membrane protein. This causes STIM1 to cluster and accumulate at ER–PM junctions. STIM1 then binds to Orai1, a Ca²⁺-permeable channel in the PM, and activates it (Lewis, 2007; Soboloff et al., 2012; Wu et al., 2014). The contributions of related proteins (Orai2, Orai3, and STIM2) to SOCE are not fully resolved (Hoth and Niemeyer, 2013), although STIM2 is usually more important than STIM1 for refilling of Ca²⁺ stores (Brandman et al., 2007). Additional proteins, including junctate, CRACR2, and SOCE-associated regulatory factor (SARAF), also interact with STIM1–Orai1 signaling complexes and regulate both activation and deactivation of SOCE (Srikanth et al., 2010, 2012, 2013; Palty et al., 2012; Srikanth and Gwack, 2012, 2013).

We show that σ1Rs constitutively inhibit SOCE and reduce the Ca²⁺ content of the ER and that σ1R ligands modulate this inhibition. The σ1R associates with STIM1 in the ER and is conveyed with STIM1 to ER–PM junctions after store depletion. This association slows the recruitment of STIM1 to the junctions and reduces binding of STIM1 to Orai1. Our results establish that σ1Rs inhibit a ubiquitous Ca²⁺ entry pathway and suggest a general model for directed translocation of σ1R to its PM targets.

SOCE in human embryonic kidney (HEK) cells can be activated by depletion of intracellular Ca²⁺ stores using thapsigargin to inhibit the ER Ca²⁺ pump or by stimuli of endogenous receptors (e.g., ATP or carbachol) that activate PLC. The contributions of Orai1 and STIM1 to SOCE (Parekh and Putney, 2005; DeHaven et al., 2009; Soboloff et al., 2012) are clear from the inhibition of thapsigargin-evoked Ca²⁺ entry in HEK cells expressing a dominant-negative form of Orai1 (Orai1^E106Q; Prakriya et al., 2006) and the enhancement of SOCE after overexpression of Orai1 with STIM1 (Fig. 1, A and B). The initial Ca²⁺ release evoked by thapsigargin was unaffected by these effects of Orai1 and STIM on SOCE. Stable expression of a V5-tagged σ1R in HEK cells (HEK-σ1R cells) attenuated the Ca²⁺ signals evoked by thapsigargin without affecting expression of Orai1 or STIM1 (103 ± 5% and 91 ± 8% of wild-type cells, respectively; Fig. 1 C) or the basal cytosolic free Ca²⁺ concentration ([Ca²⁺]_c; 45 ± 7 nM and 50 ± 3 nM in wild-type and HEK-σ1R cells, respectively). The increase in [Ca²⁺]_c after addition of thapsigargin in Ca²⁺-free medium was reduced by 65 ± 9%, and the SOCE detected after restoration of extracellular Ca²⁺ was reduced by 86 ± 4% in HEK-σ1R cells (Fig. 1, D and E). The rate of increase of [Ca²⁺]_c during SOCE decreased from 8.8 ± 0.3 nM.s⁻¹ in wild-type HEK cells to 2.8 ± 0.3 nM.s⁻¹ in HEK-σ1R cells. SOCE in HEK-σ1R cells was similarly reduced across a range of extracellular Ca²⁺ concentrations (Fig. S2 A). The inhibition of both thapsigargin-evoked Ca²⁺ release and SOCE in HEK-σ1R cells was also observed at 37°C (Fig. S2 B) and in single-cell measurements (Fig. 1 F). The diminished SOCE did not result from ineffective store emptying because it was unaffected by prolonging the incubation with thapsigargin from 10 to 20 min (Fig. 1 E). Indeed, both the initial Ca²⁺ content of the stores (determined by addition of ionomycin in Ca²⁺-free medium, the effects of which are not restricted to the ER) and the residual content after thapsigargin treatment were reduced in HEK-σ1R cells (Fig. 1, G and H). When ATP and carbachol were used to deplete Ca²⁺ stores via endogenous pathways, the Ca²⁺ release and Ca²⁺ entry were also attenuated in HEK-σ1R cells (Fig. 1, I and J). The lesser Ca²⁺ release evoked by ATP and carbachol in HEK-σ1R cells (52 ± 8% of wild-type cells) matched the reduced Ca²⁺ content of the stores (59 ± 9%), suggesting that this was responsible for the diminished response to PLC-coupled receptors.

To investigate whether sustained depletion of Ca²⁺ stores might itself cause down-regulation of SOCE, HEK cells were treated with cyclopiazonic acid (CPA) to reversibly inhibit the ER Ca²⁺ pump for a 2-h period that later experiments (see Fig. 3) show is sufficient for σ1R agonists to inhibit SOCE. This CPA treatment caused a more substantial depletion of the intracellular Ca²⁺ stores than was observed in HEK-σ1R cells, but a much smaller inhibition of the SOCE evoked by subsequent addition of thapsigargin (Fig. 2, A and B). These results establish that loss of Ca²⁺ from the ER does not cause the reduced SOCE in HEK-σ1R cells.

The smaller increase in [Ca²⁺]_c evoked by SOCE in HEK-σ1R cells could result from decreased Ca²⁺ entry or enhanced Ca²⁺ extrusion. However, rates of recovery from Ca²⁺ signals evoked by carbachol and ATP in Ca²⁺-free medium (measured over matched [Ca²⁺]_c) were unaffected by expression of σ1R (half-times, t_1/2 = 36 ± 5 s and 32 ± 6 s for wild-type and HEK-σ1R cells, respectively). The smaller [Ca²⁺]_c increases in HEK-σ1R cells were not, therefore, due to more effective buffering or Ca²⁺ extrusion. Because most Ca²⁺ extrusion pathways do not transport Mn²⁺, we used quenching of fura 2 fluorescence to measure unidirectional Mn²⁺ influx through the SOCE pathway (Fig. 2 C). Thapsigargin, or carbachol with ATP, stimulated Mn²⁺ entry in HEK cells, consistent with the activation of SOCE in response to store depletion. In HEK-σ1R cells, there was no change in the rate of Mn²⁺ entry in response to either stimulus (Fig. 2 D). Together, these results establish that stable expression of σ1R inhibits SOCE.

Selection of polyclonal HEK cells stably expressing σ1R might have propagated cells with different Ca²⁺ signaling behaviors. However, the thapsigargin-evoked increase in [Ca²⁺]_c and SOCE and the Ca²⁺ content of the intracellular stores were also reduced in HEK cells transiently expressing σ1Rs (Fig. S2, C–E). The reduced SOCE correlated with the level of expression of σ1R (Fig. S2, F and G). Translocation of GFP-tagged nuclear factor of activated T cells (NFAT) from the cytosol to the nucleus requires SOCE (Kar et al., 2011). SOCE stimulated NFAT translocation in HEK cells, and the response was attenuated to similar degrees in cells stably or transiently expressing σ1R (Fig. 2, E and F).

We conclude that expression of σ1R inhibits SOCE by reducing the coupling of empty stores to the activation of SOCE.

The σ1R agonist (+)SKF10047 (Su et al., 2010; Navarro et al., 2012) and the antagonist BD1047 (Fig. S1; Skuza and Rogóz, 2006; Gromek et al., 2014) were used to investigate the acute effects of σ1Rs in CHO cells and HEK-σ1R cells. In CHO cells, σ1Rs are endogenously expressed (Hayashi and Su, 2007). As in HEK cells, SOCE was inhibited by transient expression of Orai1^E106Q, although in CHO cells, the thapsigargin-evoked Ca²⁺ release was also inhibited (Fig. S3 A). In both CHO and HEK-σ1R cells, preincubation with BD1047 increased the amplitude of the Ca²⁺ signals evoked by SOCE, whereas the agonist (+)SKF10047 had the opposite effect (Fig. 3, A–D). Neither ligand affected SOCE in wild-type HEK cells (Fig. 3, E and F), confirming that the effects are mediated by σ1Rs. The temperature dependence and slow equilibration of ligand binding to σ1Rs (Yamamoto et al., 2001; Chu and Ruoho, 2016), together with the need to load cells with Ca²⁺ indicators at 20°C, limited opportunities to investigate the time course of the effects of σ1R ligands. Nevertheless, it is clear that treatment with ligands for at least 1 h at 37°C before loading cells with Ca²⁺ indicator (1.5 h) in the continued presence of ligands was required to detect significant effects of the ligands on SOCE (Fig. S3, B–D).

In CHO cells, siRNA to σ1R almost abolished expression of endogenous σ1R, but this was accompanied by reduced expression of Orai1 and increased expression of STIM1 (Fig. 3, G and H). The loss of Orai1 could reflect a chaperone role for σ1R similar to the stabilization of human ether-a-go-go–related gene (HERG) K⁺ channels by σ1R (Hayashi and Su, 2007; Crottès et al., 2011). Alternatively, Orai1 expression may be down-regulated through an adaptive feedback mechanism arising from the reduced inhibition of SOCE after loss of σ1Rs. Overstimulation of SOCE by constitutively active STIM1 was shown previously to reduce Orai1 expression (Kilch et al., 2013). Despite the reduced expression of Orai1, SOCE was increased in CHO cells lacking σ1Rs, and it was then unaffected by (+)SKF10047 or BD1047 (Fig. 3, I and J). The enhanced SOCE in CHO cells lacking σ1Rs was abolished by expression of the dominant-negative form of Orai1 (Fig. S3 G), confirming that it was mediated by Orai1. SOCE monitored by unidirectional Mn²⁺ entry was also increased in CHO cells treated with siRNA to σ1R (Fig. S3, E and F). In normal CHO cells, (+)SKF10047 reduced the Ca²⁺ content of the stores, whereas the σ1R antagonist BD1047 increased their content to a level that matched that of cells without σ1Rs. Neither ligand affected the Ca²⁺ stores in CHO cells lacking σ1Rs (Fig. 3, K and L). Although comparison of SOCE-mediated Ca²⁺ signals in CHO cells with and without σ1Rs is compromised by accompanying changes in STIM1 and Orai1 expression (Fig. 3, G and H), the analyses demonstrate that σ1R ligands are effective only in cells expressing σ1Rs, and they establish a constitutive inhibition of SOCE by endogenous σ1Rs and an associated reduction in ER Ca²⁺ content in CHO cells (Fig. 3, I-L). Similar results were obtained in HEK-σ1R cells: siRNA to σ1R abolished the effects of σ1R ligands on both SOCE and the Ca²⁺ content of the stores; it also increased the basal Ca²⁺ content of the stores and the rate of Mn²⁺ entry evoked by either thapsigargin or by the more physiological stimuli, ATP and carbachol (Fig. S3, H and J).

Determining whether ligands of σ1R are more effective before or after depletion of Ca²⁺ stores was frustrated by the need for prolonged preincubations at 37°C for optimal effects (Fig. S3, B–D). In a modified protocol, fluo 4–loaded HEK-σ1R cells in Ca²⁺-free Hepes-buffered saline (HBS) were incubated with (+)SKF10047 or BD1047 for 2 h at 20°C, with thapsigargin added either immediately before the ligands or after the 2-h incubation. Under these conditions, where the effects of the ligands were much reduced, (+)SKF10047 modestly inhibited SOCE, and BD1047 modestly enhanced SOCE, but only when added before store depletion (Fig. S3 K). These results suggest that σ1R ligands affect an early step in the activation of SOCE.

Breast cancer cells express high levels of σ1Rs (Spruce et al., 2004; Wang et al., 2004; Aydar et al., 2006). In MDA-MB-231 human breast cancer cells, which also express σ1Rs, SOCE was enhanced by BD1047 and inhibited by (+)SKF10047 (Fig. S4). The Ca²⁺ content of the stores was also reduced by (+)SKF10047. Hence, in three cell types, HEK-σ1R, CHO, and MDA-MB-231 cells, σ1Rs both inhibit SOCE and decrease the Ca²⁺ content of the ER. The inverse agonist effect of BD1047 in CHO and MDA-MB-231 cells suggests a constitutive regulation of SOCE and ER Ca²⁺ content by endogenous σ1Rs.

Interactions between σ1R and STIM1 in unstimulated cells were investigated using HEK cells transiently expressing STIM1-Myc and σ1R-FLAG. Anti-Myc beads pulled down σ1R-FLAG from solubilized cell extracts, but only in cells expressing STIM1-Myc. Conversely, anti-FLAG beads pulled down STIM1-Myc, but only in cells expressing σ1R-FLAG (Fig. 4, A and B). Coimmunoprecipitation of σ1R-FLAG with STIM1-Myc was enhanced by (+)SKF10047 and reduced by BD1047 (Fig. 4, C–E). These results show that STIM1 and σ1R are associated in unstimulated cells and that their interaction is regulated by σ1R ligands. Furthermore, the increased association of σ1R with STIM1 evoked by the σ1R agonist (Fig. 4) correlates with the inhibition of SOCE (Fig. 3).

To investigate the intracellular dynamics of σ1R and STIM1, we used HeLa cells because they are better suited than HEK cells for optical analyses of ER proteins while still lacking detectable endogenous σ1Rs (Fig. 5 A). In cells expressing σ1R-EGFP with mCh-STIM1, σ1R-EGFP and mCh-STIM1 colocalized within the ER (Mander’s correlation coefficient was 0.77 ± 0.03; n = 8; Fig. 5 B). We used total internal reflection fluorescence (TIRF) microscopy to visualize translocation of mCh-STIM1 and σ1R-EGFP in response to thapsigargin. In cells expressing mCh-STIM1, thapsigargin stimulated an accumulation of mCh-STIM1 in puncta immediately beneath the PM (Fig. 5 C, top). This is consistent with evidence that store depletion causes STIM1 to aggregate into sub-PM clusters, where they interact with Orai1 to activate SOCE (Liou et al., 2007; Wu et al., 2014). In contrast, thapsigargin had no detectable effect on the sub-PM distribution of σ1R-EGFP expressed alone (Fig. 5 C, bottom). However, when mCh-STIM1 and σ1R-EGFP were coexpressed, thapsigargin caused both proteins to accumulate in sub-PM puncta, within which the proteins colocalized (Mander’s correlation coefficient was 0.77 ± 0.04; n = 8; Fig. 5 D), but expression of σ1R slowed the rate of formation of the mCh-STIM1 puncta (Fig. 5 E). Rates of formation of mCh-STIM1 puncta were unaffected by expression of another ER membrane protein, IP₃R1 (times to 50% accumulation were 325 ± 14 s and 342 ± 11 s, with and without IP₃R1, respectively), confirming that the effects of σ1R were not caused by nonspecific accumulation of ER proteins. Furthermore, in cells coexpressing Orai1-EGFP, σ1R-mKate, and HA-STIM1, Orai1-EGFP and σ1R-mKate accumulated into colocalized puncta after thapsigargin treatment, but neither formed puncta in the absence of STIM1 (Fig. 5, F and G). These results demonstrate that after store depletion, σ1R accompanies STIM1 to ER–PM junctions, but σ1R slows the accumulation of STIM1.

In related experiments, HeLa cells expressing different combinations of σ1R-EGFP, mCh-STIM1, and Orai1-Myc were fixed for immunolabeling, and confocal images were analyzed to assess colocalization of the proteins before and after treatment with thapsigargin. As expected, in cells coexpressing STIM1 and Orai1, thapsigargin caused their colocalization to increase, consistent with evidence that clustered STIM1 at ER–PM junctions captures Orai1 as it diffuses within the PM (Wu et al., 2014). In contrast, when σ1R and Orai1 were coexpressed, their colocalization was enhanced by store depletion only in the presence of STIM1, and overlapping puncta of all three proteins were then apparent at the cell periphery (Fig. 5, H and I). These results agree with those obtained using TIRF microscopy and demonstrate the importance of STIM1 in recruiting both Orai1 and σ1R to the same junctions.

The requirement for STIM1 in recruiting σ1R to ER–PM junctions containing Orai1 was investigated further by expressing σ1R-FLAG and Orai1-Myc with and without HA-STIM1. After treatment with thapsigargin and cell surface biotinylation, PM protein complexes were purified using avidin. Immunoblotting showed that the amount of σ1R within the biotinylated sample was significantly increased in cells overexpressing STIM1 and Orai1, but not when only Orai1 was overexpressed (Fig. 6, A and B). This indicates that STIM1 either promotes trafficking of σ1R to the PM, where it is directly biotinylated, or it promotes association of σ1R with a biotinylated PM protein. Similar analyses established that expression of σ1R-FLAG reduced the amount of STIM1 in the biotinylated sample to 47 ± 12% (n = 3) of that measured without σ1R (Fig. 6, C and D). The β-actin control showed no evidence of cell permeabilization or biotinylation of intracellular proteins. The biotinylated PM sample was subjected to a further round of purification using anti-Myc beads. Immunoblotting confirmed that when all three proteins were expressed, they were each captured in the final extract, suggesting that both STIM1 and σ1R are associated with the PM Orai1 channel complex (and that there is no need to invoke cell surface expression of σ1R to account for its presence in the biotinylated sample). The amount of STIM1 within this complex was again reduced by σ1R to 51 ± 7% of that measured without σ1R (Fig. 6, C and D). These results indicate that σ1R reduces the amount of STIM1 bound to PM Orai1. This was confirmed by purifying HA-STIM1 with anti-HA beads: the amount of Orai1 that copurified with STIM1 was reduced in the presence of σ1R (Fig. 6 E).

If the reduction in STIM1 binding to Orai1 contributes to inhibition of SOCE by σ1R, we might expect increased expression of STIM1 to relieve the inhibition. We therefore tested the effects of overexpressing STIM1 on the amplitude of SOCE in wild-type HEK and HEK-σ1R cells. Expression of STIM1 produced a similar increase in the amplitude of SOCE in wild-type and HEK-σ1R cells, but the percent increase was greater in the HEK-σ1R cells (36 ± 5% in wild-type and 81 ± 8% in HEK-σ1R cells; Fig. 6 F). This suggests that activation of SOCE is more limited by STIM1 in HEK-σ1R cells than in wild-type cells. The effects of STIM1 on SOCE were matched by its effects on Ca²⁺ stores: overexpression of STIM1 increased the Ca²⁺ content of the stores, and the effect was greater in HEK-σ1R relative to wild-type cells (83 ± 7% and 18 ± 5% increases, respectively; Fig. 6 G). Furthermore, the effects of σ1R ligands on SOCE were much reduced in HEK-σ1R cells overexpressing STIM1 and Orai1 (Fig. S5). These results support the idea that σ1Rs inhibit the association of STIM1 with PM Orai1, thereby reducing SOCE (Fig. 6 H). Coincident with this inhibition of SOCE by σ1Rs, we invariably detected a decrease in the Ca²⁺ content of the ER.

To examine the structure of the PM Orai1 channel complex in the presence and absence of σ1R, we used atomic force microscopy (AFM). Previous AFM images of complexes purified from cells overexpressing Orai1 and STIM1 showed a hexameric arrangement of STIM1 around a central Orai1 complex and a few strings of STIM1 molecules associated with Orai1, consistent with the oligomerization of STIM1 after depletion of Ca²⁺ stores (Balasuriya et al., 2014b). We examined extracts from thapsigargin-treated HEK cells expressing Orai1-Myc/His and HA-STIM1, with or without σ1R-FLAG, in which cell surface proteins had been biotinylated and complexes were isolated by sequential purification using avidin and anti-Myc beads. AFM images showed large particles decorated by smaller peripheral particles (Fig. 7 A). The large central particle had the volume expected of hexameric Orai1 (566 ± 8 nm³; Fig. 7 B). A volume distribution of bound peripheral particles for the Orai1-Myc/HA-STIM1 sample had two peaks at 131 ± 2 and 235 ± 4 nm³ (Fig. 7 C), consistent with the expected volumes of STIM1 monomers and dimers. For the Orai1-Myc/HA-STIM1/σ1R-FLAG sample, the volume distribution of the peripheral particles had three peaks (62 ± 13, 130 ± 20, and 220 ± 22 nm³) corresponding to STIM1 monomers and dimers and a smaller peak consistent with the expected volume of σ1R monomers (∼63 nm³; Fig. 7 D). Of the 300 Orai1 complexes analyzed when expressed with STIM1 alone, 73 had bound particles and were either singly or doubly decorated. The total number of bound STIM1 was 96. From the 300 Orai1 complexes analyzed when coexpressed with STIM1 and σ1R, 76 had bound particles; there were 59 bound STIM1 and 46 bound σ1R. So the total number of bound STIM1 was reduced by 39% in the presence of σ1R. AFM images of Orai1 isolated from cells expressing Orai1 and STIM1 revealed, albeit with low frequency, that Orai1 bound to strings of STIM1 (Fig. 7 E). These assemblies were never seen in images from cells coexpressing σ1R. These results provide evidence for a PM complex of Orai1, STIM1, and σ1R and for reduced binding of STIM1 to Orai1 in the presence of σ1R.

Reduced binding of STIM1 to Orai1 caused by σ1R is expected to reduce SOCE, but σ1R might also directly inhibit gating of Orai1 channels. To address this possibility, we used the channel-activating domain (CAD) of STIM1, which directly activates Orai1 (Muik et al., 2009; Park et al., 2009; Yuan et al., 2009; Gudlur et al., 2014). mCh-CAD expressed alone in HeLa cells was diffusely distributed in the cytoplasm, but it was peripherally distributed when coexpressed with Orai1, consistent with the constitutive association of CAD and Orai1 (Fig. 8 A). Addition of extracellular Ca²⁺ to HEK or HEK-σ1R cells in Ca²⁺-free HBS had no significant effect on [Ca²⁺]_c, but there was a substantial increase in [Ca²⁺]_c in cells expressing CAD, consistent with constitutive activation of SOCE by CAD (Fig. 8, B and C). The response was indistinguishable in HEK and HEK-σ1R cells, suggesting that σ1R does not directly modulate PM expression of Orai1 nor its activity.

We have shown that σ1Rs inhibit SOCE by decreasing the effectiveness with which empty stores stimulate Orai1. The target for regulation of SOCE by σ1R appears to be STIM1 (Fig. 8 D). σ1R and STIM1 colocalize in the ER; they can be coimmunoprecipitated before and after depletion of Ca²⁺ stores, and their interaction is regulated by σ1R ligands. The agonist, (+)SKF10047, increases binding of STIM1 to σ1R and further inhibits SOCE, whereas the antagonist, BD1047, has the opposite effects. After store depletion, σ1R translocates with STIM1 to ER–PM junctions, but σ1R slows recruitment of STIM1 and reduces the amount of STIM1 bound to PM Orai1. This reduction in STIM1 binding to Orai1 suggests a likely mechanism for the inhibition of SOCE wherein σ1R accompanies STIM1 to ER–PM junctions, where it attenuates the interaction of STIM1 with Orai1. The gap between the ER and PM at the junctions where SOCE occurs is probably too large (>9 nm; Várnai et al., 2007) to be bridged by the short cytosolic loop of σ1R (Fig. S1; Hayashi and Su, 2007). The association of σ1Rs with PM Orai1 is therefore likely to be mediated by STIM1. Reduced binding of STIM1 to Orai1 in the presence of σ1R may be caused by σ1R inhibiting the oligomerization of STIM1 or directly reducing the affinity of STIM1 for Orai1.

There are interesting similarities between the behavior of σ1Rs and that of other ER membrane proteins, including SARAF (Palty et al., 2012) and POST (partner of STIM1; Krapivinsky et al., 2011). SARAF also translocates to ER–PM junctions in a STIM1-dependent manner, and it promotes deactivation of STIM1 by antagonizing interactions between STIM1 molecules (Palty et al., 2012). Translocation of POST modulates SOCE-evoked Ca²⁺ signals because it inhibits the PM Ca²⁺ pump that extrudes cytosolic Ca²⁺ (Krapivinsky et al., 2011). Hence, after loss of Ca²⁺ from the ER, STIM1 both activates SOCE and fine-tunes its activity by delivering additional Ca²⁺-regulating proteins to ER–PM junctions (Fig. 8 D). For σ1Rs, the effects of ER luminal Ca²⁺ on these delivery mechanisms may operate at two levels. Loss of ER Ca²⁺ (or a σ1R agonist) releases σ1R from its interaction with the ER luminal protein, BiP (Fig. S1; Hayashi and Su, 2007). Store depletion also causes STIM1 to oligomerize and thereby gain affinity for ER–PM junctions. Depletion of Ca²⁺ stores may therefore both release σ1R from its ER tethers and, via its association with oligomeric STIM1, allow it to accumulate at ER–PM junctions. We focused on SOCE, but recruitment of σ1Rs to ER–PM junctions by STIM1 might also be involved in regulation of other PM channels by σ1Rs (Maurice and Su, 2009; Su et al., 2010; Kourrich et al., 2013; Pabba, 2013). For example, the L-type Ca²⁺ channel is inhibited by σ1R (Tchedre et al., 2008) and by depletion of intracellular Ca²⁺ stores and STIM1 (Park et al., 2010; Wang et al., 2010). We suggest that STIM1-mediated translocation of σ1R to ER–PM junctions may inhibit voltage-gated Ca²⁺ entry and may also deliver σ1Rs to additional PM targets (Fig. 8 D).

Inhibition of SOCE by σ1Rs was invariably accompanied by a decrease in the Ca²⁺ content of the ER with no evident change in [Ca²⁺]_c. In contrast, and consistent with a study by López et al. (2012), inhibition of SOCE by expression of Orai1^E106Q did not consistently affect ER Ca²⁺ content: it was normal in HEK cells but reduced in CHO cells. Inhibition of the STIM1–Orai1 interactions that mediate thapsigargin-evoked SOCE are not, therefore, sufficient to explain the effects of σ1Rs on ER Ca²⁺ content. It may be that σ1Rs also interact with STIM2, which plays a major role in maintaining the Ca²⁺ content of the stores (Brandman et al., 2007), or with other proteins, such as the sarco/ER Ca²⁺ ATPase, as was shown for orosomucoid-like 3 (Cantero-Recasens et al., 2010), or with Ca²⁺ channels that mediate Ca²⁺ uptake and release from the ER. For example, Sec61 mediates Ca²⁺ release from the ER, and it is inhibited by BiP (Schäuble et al., 2012). Expression of σ1R might sequester BiP (Fig. S1) and thereby enhance the Sec61-mediated Ca²⁺ leak. The decreased Ca²⁺ content of the ER might also arise from σ1R stabilizing IP₃R3 and thereby enhancing Ca²⁺ transport from the ER to mitochondria (Hayashi and Su, 2007).

The pathophysiological effects σ1Rs may, in part, result from inhibition of SOCE and the reduced Ca²⁺ content of the ER. The latter may affect protein folding (Hayashi and Su, 2007) and inhibit apoptosis by preventing excessive Ca²⁺ transfer to mitochondria (Maurice and Su, 2009; Giorgi et al., 2012). The effects of σ1Rs on mitochondrial Ca²⁺ uptake are probably finely balanced because σ1Rs enhance delivery of Ca²⁺ to mitochondria at MAMs by stabilizing MAM-associated IP₃Rs (Hayashi and Su, 2007), whereas our results show that σ1Rs reduce the ER Ca²⁺ content. The latter could explain the otherwise surprising antiapoptotic effects of σ1Rs (Wang et al., 2005; Maurice and Su, 2009; Decuypere et al., 2011; Crottès et al., 2013). The σ1R agonist, cocaine, was recently shown to attenuate SOCE in rat brain microvascular endothelial cells (Brailoiu et al., 2016). The neuroprotective effects of σ1R agonists after ischemic injury (Katnik et al., 2006) and in patients with amyotrophic lateral sclerosis arising from loss-of-function mutations in σ1R (Al-Saif et al., 2011; Ono et al., 2014) may also, at least in part, be due to inhibition of SOCE. Hyperactive SOCE may contribute to the motor deficiencies in σ1R-knockout mice (Maurice and Su, 2009; Sabino et al., 2009; Mavlyutov et al., 2010) and to neurodegeneration in Alzheimer’s (Mishina et al., 2008; Ishikawa and Hashimoto, 2009; Hyrskyluoto et al., 2013) and Parkinson’s diseases (Mishina et al., 2005; Hyrskyluoto et al., 2013; Francardo et al., 2014), where expression of σ1R is reduced. These suggestions prompt consideration of whether σ1R also interacts with STIM2 because it appears to play the major role in regulating SOCE in central neurons (Berna-Erro et al., 2009).

We conclude that σ1Rs inhibit SOCE because they associate with STIM1, slow STIM1 recruitment to ER–PM junctions, and reduce its binding to Orai1 after depletion of Ca²⁺ stores. Our study highlights a role for STIM1 in translocating σ1Rs to the PM and establishes σ1Rs and their ligands as important regulators of SOCE, a ubiquitously expressed Ca²⁺ entry pathway (Fig. 8 D).

(+)SKF10047 and BD1047 were from Tocris Bioscience. Ionomycin was from MerckEurolab. Thapsigargin was from Alomone Labs. Anti-Myc monoclonal antibody (1:500 dilution for immunoblots; 46–0603), fura 2–AM, and fluo 4–AM were from Thermo Fisher Scientific. Anti-HA (1:500; 16B12) and anti-FLAG (1:500; F3165) monoclonal antibodies were from Covance and Sigma-Aldrich, respectively. The anti-σ1R antibody (1:200; Ab53852), which recognizes a sequence conserved in human and mouse σ1R, was from Abcam. Custom-made rabbit polyclonal antipeptide antisera to STIM1 (1:100; CDPQHGHGSQRDLTR; the Cys used for conjugation is underlined) and Orai1 (1:200; CEFAWLQDQLDHRGD) were prepared by Sigma-Aldrich. Anti-actin (1:500; A5441) antibody was from Sigma-Aldrich. Anti–mouse (1:1,000) and anti–rabbit (1:1,000) HRP-conjugated secondary antibodies were from Thermo Fisher Scientific and Bio-Rad Laboratories, respectively. Sources of additional materials are provided within the relevant methods.

Plasmids encoding HA-STIM1 and Orai1-Myc/His₆ have been described previously (Willoughby et al., 2012; Balasuriya et al., 2014b). For mCh-STIM1, human STIM1 was subcloned into mCherry-C1 (Takara Bio Inc.) using XbaI and NotI. For σ1R-FLAG, σ1R was subcloned into pcDNA3.1/FLAG using HindIII and AgeI. For σ1R-GFP, σ1R was subcloned into GFP-N1 (Takara Bio Inc.) using HindIII and KpnI. For σ1R-V5, σ1R was subcloned into pcDNA3.1/V5-His-TOPO using HindIII and AgeI. For σ1R-mKate, σ1R was subcloned into mKate2-N (Evrogen) using HindIII and KpnI. The coding sequences of all new constructs were verified. pDsRed2-Mito was from Takara Bio Inc. A pSIREN vector encoding siRNA for σ1R (5′-GATCCACACGTGGATGGTGGAGTATTCAAGAGATACTCCACCATCCACGTGTTTTTTTGCTAGCG-3′) was used to inhibit expression of σ1R. pSIREN encoding the luciferase gene was used as a negative control. Both pSIREN constructs were gifts from T.-P. Su (National Institutes of Health, Bethesda, MD; Hayashi and Su, 2004). An expression plasmid (MO70) encoding a dominant-negative form of Orai1 in which Glu-106 is replaced by Gln (Orai1^E106Q) was a gift from Y. Gwack and S. Srikanth (University of California, Los Angeles, Los Angeles, CA; Srikanth et al., 2012). The expression plasmid for mouse GFP-NFAT1 was a gift from A. Parekh (University of Oxford, Oxford, England, UK; Kar et al., 2011). The mCh-STIM1 CAD expression plasmid was a gift from P. Hogan (La Jolla Institute for Allergy and Immunology, La Jolla, CA; Gudlur et al., 2014), and Orai1-CFP was from D.M.F. Cooper (University of Cambridge, Cambridge, England, UK).

All cells were maintained in DMEM with 10% fetal bovine serum and 1% penicillin/streptomycin (Sigma-Aldrich) at 37°C in humidified air with 5% CO₂. tsA 201 cells were grown to 70% confluence in a 162-cm² flask and transfected using calcium phosphate. 50 µg of plasmid DNA was mixed with 5 ml of 250 mM CaCl₂ and diluted with 5 ml of medium comprising 275 mM NaCl, 10 mM KCl, 1.4 mM Na₂HPO₄, 15 mM glucose, and 42 mM Hepes, pH 7.07. The mixture was added to the cells bathed in 25 ml of fresh growth medium. After 8 h, the medium was replaced with fresh growth medium. Cells were incubated for a further 48 h before being used for experiments.

HEK 293 cells were transfected using polyethylenimine. For cells grown to 70% confluence in 1 well of a 6-well plate, 1 µg of plasmid DNA was mixed with 2 µl of 7.5 mM polyethylenimine (Polysciences, Inc.) and then diluted with 150 µl of serum-free DMEM. The mixture was incubated for 10 min at 20°C and then added to wells containing 2 ml of fresh growth medium for 48 h. The generation of a polyclonal HEK cell line stably expressing mouse σ1R-V5 (HEK-σ1R cells) was performed as described previously (Xu et al., 2012). These cells were maintained in medium supplemented with 0.8 mg/ml G418 (Thermo Fisher Scientific).

HeLa cells were grown on poly-l-lysine–coated 25-mm glass coverslips and transfected using Lipofectamine 2000 (Thermo Fisher Scientific). For 1 well of a 6-well plate, 2 µg of plasmid DNA was diluted in 200 µl Opti-MEM and incubated at 20°C for 5 min. This was combined with 200 µl Opti-MEM containing 4 µl Lipofectamine 2000 and left for a further 20 min at 20°C. The mixture was then added to cells in 2 ml of fresh medium. Cells were incubated for 48 h at 37°C and then used for experiments.

For measurements of [Ca²⁺]_c in populations of cells, HEK cells were seeded into poly-l-lysine–coated 96-well plates. After 24 h, cells were incubated with 2 µM fluo 4–AM in HBS for 60 min at 20°C. HBS had the following composition: 150 mM NaCl, 10 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 10 mM glucose, and 10 mM Hepes, pH 7.3. In Ca²⁺-free HBS, Ca²⁺ was omitted, and 1 mM BAPTA was added. For treatments with (+)SKF10047 and BD1047, 10 mM stock solutions were prepared in DMSO and water, respectively. Cells were pretreated with 25 µM (+)SKF10047 or 10 µM BD1047 in serum-free DMEM for 1 h at 37°C before loading cells with fluorescent Ca²⁺ indicators. Drug treatments were continued during loading and throughout [Ca²⁺]_c measurements. After loading, cells were washed and incubated in HBS for a further 30 min at 20°C. Fluorescence (excitation at 490 nm and emission at 520 nm) was measured at 20°C using a plate reader that allows online additions (FlexStation 3; Molecular Devices). Fluorescence was calibrated to [Ca²⁺]_c from

where K_D = 345 nM (K_D = 190 nM at 37°C; Fig. S2 B), F is the measured fluorescence, and F_max and F_min are the fluorescence values determined after addition of 0.1% Triton X-100 in HBS with 10 mM Ca²⁺ or 10 mM BAPTA, respectively.

Confluent cultures of HEK cells in 96-well plates were loaded with 2 µM fura 2–AM using the method described for fluo 4. Fluorescence (excitation at 360 nm and emission at 510 nm) was measured using a plate reader (FlexStation 3) at 1.5-s intervals at 20°C. Quenching of fura 2 fluorescence (which reports unidirectional entry of Mn²⁺) is reported as F/F₀, where F is the fluorescence intensity recorded at each time and F₀ is the mean fluorescence intensity measured in the 5 s before addition of MnCl₂. Monoexponential curve fits to the time course of the changes in F/F₀ were used to compute half-times (t_1/2) for Mn²⁺-evoked fluorescence quenching.

HEK cells were seeded onto poly-l-lysine–coated 25-mm coverslips, transfected with GFP-NFAT plasmid using polyethylenimine, and used after 48 h. The distribution of GFP fluorescence was measured before and 40 min after addition of 5 µM thapsigargin to cells at 37°C in HBS. Fluorescence (excitation at 488 nm and emission at 510–540 nm) was collected using a confocal microscope (SP5; Leica Biosystems) with an oil-immersion 40× objective (NA 1.25). Analyses of nuclear translocation of GFP-NFAT were performed with coded images, which were decoded only when the analysis was complete.

Cells were grown in 162-cm² flasks. Where appropriate, cells were transfected with 50 µg of plasmid DNA using polyethylenimine. Cells were extracted in ice-cold medium (138 mM NaCl, 5 mM KCl, 1 mM Na₂HPO₄, 7.5 mM glucose, 21 mM Hepes, and 2 mM EDTA, pH 7.4) and centrifuged at 1,000 g for 5 min. Pelleted cells were solubilized at 4°C for 60 min in Triton solution (TS) containing 25 mM Tris-HCl, 150 mM NaCl, 10 mM EDTA, 1% Triton X-100, and 1 mg/ml protease inhibitor cocktail solution (Roche), pH 7.4, and samples were analyzed by SDS-PAGE followed by immunoblotting.

tsA 201 cells, which are SV40-transformed HEK 293 cells, were used because they express heterologous proteins at high levels. Cells were grown in 162-cm² flasks and transfected using calcium phosphate. Pretreatments with (+)SKF10047 and BD1047 were for 2 h at 37°C, and stimulation with 5 µM thapsigargin was for 30 min at 20°C. Cells were extracted in 25 ml of ice-cold medium, and all subsequent steps were performed at 4°C. The suspension was centrifuged at 1,000 g for 5 min, and pelleted cells were solubilized for 60 min in 500 µl TS. After centrifugation (50,000 g for 60 min), 50 µl of the supernatant was removed for analysis of total expression (input), and 450 µl was incubated with 30 µl anti-Myc (EZ View Red) or anti-FLAG beads (Sigma-Aldrich) for 3 h with rotation. Protein–bead complexes were isolated (20,800 g for 10 min) and washed three times in TS, and proteins were eluted either with 50 µl of the peptides (1 mg/ml; Sigma-Aldrich), to which the anti-Myc or anti-FLAG antibodies had been raised, or with 50 µl Laemmli buffer. The eluted samples were analyzed by SDS-PAGE followed by immunoblotting. For immunoblots, lanes were loaded with 10 µl of the 500-µl sample (2% of the entire sample) for the measurement of input and with 10 or 20 µl of the 50-µl eluate for measurements of immunoprecipitation.

tsA 201 cells were grown in 162-cm² flasks and transfected using calcium phosphate. After appropriate stimulation, the medium was removed and replaced with 12.5 ml of ice-cold HBS containing 0.2 mg/ml biotin-sulfo-NHS (Thermo Fisher Scientific). After 60 min on ice, cells were washed three times with 15 ml Tris-buffered saline (25 mM Tris-HCl, 150 mM NaCl, and 10 mM EDTA, pH 7.4) and centrifuged at 1,000 g for 5 min, and the pellet was solubilized in 500 µl TS for 60 min at 4°C. After centrifugation at 50,000 g for 60 min, the supernatant was incubated with 50 µl monomeric avidin-coated agarose beads (Thermo Fisher Scientific) at 4°C for 2 h. Protein–bead complexes were collected at 20,800 g for 10 min, washed three times in TS, and eluted with either 50 µl Laemmli buffer for immunoblots or biotin (1 mg/ml in 1 ml TS) for further immunopurification using anti-Myc beads as described in the previous paragraph. For analyses of avidin pull-downs of biotinylated proteins (Fig. 5, C–E), 2% of the total sample was loaded as input, and 40% of the Laemmli sample was loaded in the surface biotinylation lanes.

HeLa cells were seeded on poly-l-lysine–coated glass coverslips, transfected, and used after 48 h. After stimulation, cells were washed with ice-cold PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 2 mM KH₂PO₄), fixed with 4% paraformaldehyde for 20 min, and permeabilized with 0.5 mg/ml saponin for 60 min (Sigma-Aldrich) in blocking solution (5% goat serum and 3% BSA in PBS). Cells were stained with primary antibody in blocking solution (PBS containing 3% BSA and 5% goat serum) for 60 min at 20°C, washed twice with PBS, and then incubated in the dark with secondary antibody in blocking solution for 60 min at 20°C, washed with PBS, dried, mounted onto a glass microscope slide, and stored at 4°C. Cells were imaged using an oil-immersion 60× objective (NA 1.40) using a confocal microscope (SP5; Leica Biosystems). For both Pearson’s and Mander’s coefficient measurements, images were analyzed with ImageJ (National Institutes of Health) using the JACoP plugin. For Mander’s coefficient, only pixels in which HA-STIM1 (or σ1R-EGFP) was detected were considered, and the fraction of those pixels in which Orai1-Myc was also detected was then computed to provide the colocalization coefficient.

Coverslips were mounted on a TIRF microscope (IX51 inverted microscope [Olympus] with a 100× oil-immersion objective [NA 1.49] coupled to an electron-multiplying charged-coupled device camera [iXon; Andor Technology] and 488-nm argon ion and 561-nm diode lasers). Cells were incubated with HBS at 20°C and imaged (1 image/s) by exciting σ1R-GFP at 488 nm (emission at 510–540 nm) and mCh-STIM1 at 561 nm (emission at 610–650 nm). For each experiment, there were suitable controls, with cells expressing the EGFP-tagged protein alone and the mCherry/mKate-tagged protein alone to ensure there was no bleed through. For depletion of stores, cells were incubated with 1 µM thapsigargin in Ca²⁺-free HBS. Fluorescence intensities were quantified using the time series analyzer plugin V2.0 in ImageJ. Individual regions of interest within the cell were selected, and the data were analyzed as F/F₀, where F and F₀ are the fluorescence intensities at each time and at the start of the experiment, respectively.

tsA 201 cells expressing appropriate combinations of Orai1-Myc-His, σ1R-FLAG, and HA-STIM1 were treated with thapsigargin, followed by biotin-sulfo-NHS, and then purified using sequential avidin and anti-Myc affinity chromatography, as described in the Isolation of surface biotinylated proteins section. About 45 µl of proteins was added to a 1-cm² mica disk, incubated at 20°C for 10 min, gently washed with water, and dried under nitrogen. Samples were imaged in air using an atomic force microscope (Multimode; Bruker). The silicon cantilever (OTESPA; Bruker) was set at a drive frequency of 271–321 kHz and spring constant of 12–103 N/m. The scan rate was 3 Hz, and the applied imaging force was kept as low as possible (target amplitude of 1.0 V and amplitude set point of 0.7–1.0 V). Molecular volumes for individual particles were determined using an image processor (version 5; Scanning Probe). For particles within complexes, particle heights (h) and radii (r) were measured manually using Nanoscope software. Particle volumes (V_m) were then calculated from

Molecular volume (V_c), based on a known molecular mass (M₀), was calculated from

where N₀ is Avogadro’s number, V₁ is the specific particle volume (0.74 cm³/g), V₂ is the water specific volume (1 cm³/g), and d is the extent of hydration (assumed to be 0.4 g H₂O/g protein).

Most results are presented as mean ± SEM from n independent experiments. Statistical analysis used Student’s t test or analysis of variance (ANOVA) followed by Tukey’s posthoc test as appropriate.

Fig. S1 illustrates key features of σ1R, and Table S1 describes ligands targeting σ1R. Fig. S2 shows the effects of expressing σ1Rs in HEK cells on SOCE and the Ca²⁺ content of the intracellular stores. Fig. S3 shows the effects of σ1R ligands on SOCE in CHO and HEK cells. Fig. S4 shows the effects of σ1R ligands on SOCE and the Ca²⁺ content of the intracellular stores in MDA-MB-231 human breast cancer cells. Fig. S5 shows the effects of σ1R ligands on SOCE in HEK-σ1R cells overexpressing STIM1 and Orai1.

We thank T.-P. Su for σ1R siRNA and control plasmids, S. Srikanth and Y. Gwack for the Orai1^E106Q construct, A. Parekh for the GFP-NFAT plasmid, P. Hogan for the CAD construct, and D.M.F. Cooper for the Orai1-CFP construct.

S. Srivats is supported by the Cambridge International and European Trust, D. Balasuriya is supported by a David James bursary from the Department of Pharmacology, University of Cambridge, and G. Vistal is supported by the Jardines Matheson student bursary. This work was supported by grants to J.M. Edwardson and R.D. Murrell-Lagnado from the Biotechnology and Biological Sciences Research Council (BB/J018236/1 and BB/F001320/1) and by a grant from the Wellcome Trust (101844) to C.W. Taylor.

The authors declare no competing financial interests.