The endocannabinoid N-arachidonoyl glycine (NAGly) inhibits store-operated Ca²⁺ entry by preventing STIM1–Orai1 interaction

Endocannabinoids are signaling lipids that modulate a multitude of physiological functions. Initially, N-arachidonoylethanolamid, also referred to as anandamide (AEA), and 2-aracidonoylglycerol (2-AG) were identified as natural ligands of the cannabinoid receptors CB₁ and CB₂ and their psychoactive effects in the central nervous system (CNS) were established (Pertwee, 2006; Mechoulam, 2002). However, evidence accumulated recently that the actions of these endocannabinoids and many other structurally related compounds are not restricted to the CNS. In particular, endocannabinoids have been implicated in the regulation of the immune system (Tanasescu and Constantinescu, 2010), the gastrointestinal tract (Izzo et al., 2001), and the cardiovascular system (Kunos et al., 2002). Interestingly, the broad spectrum of physiological functions modulated by endocannabinoids is only partially based on their binding to the classical cannabinoid receptors CB₁ and CB₂. Additional receptors such as the ‘orphan’ G-protein-coupled receptors GPR18 (Kohno et al., 2006) and GPR55 (Waldeck-Weiermair et al., 2008) have also been identified as targets of endocannabinoids. Moreover, recent studies indicate that endocannabinoids directly modulate the activity of diverse ion channels. Effects that are, at least in part, independent from G-protein-coupled receptors include the modulation of Ca²⁺-activated K⁺ channels (Parmar and Ho, 2010; Bondarenko et al., 2011a), the inhibition of voltage-gated Ca²⁺ channels (Barbara et al., 2009), and the activation of non-selective cation channels (Zygmunt et al., 1999). These receptor-independent actions of endocannabinoids have been described in human endothelial cells (Waldeck-Weiermair et al., 2008; Bondarenko et al., 2011b) and point to the capability of these lipid mediators to modulate the cellular ion homeostasis of non-excitable cells.

In this particular cell type, the major Ca²⁺ entry pathway is the so-called store-operated Ca²⁺ entry (SOCE) (Parekh and Putney, 2005). The activation of SOCE is triggered by Ca²⁺ depletion of the endoplasmic reticulum (ER), whereupon the decrease of the ER Ca²⁺ concentration ([Ca²⁺]_ER) is sensed by the stromal interacting molecule 1 (STIM1) (Roos et al., 2005; Liou et al., 2005), which oligomerizes and redistributes to subplasmalemmal ER domains, forming so-called STIM1 clusters (Wu et al., 2006; Luik et al., 2008; Liou et al., 2007). Subsequently, the interaction of STIM1 with the SOCE pore-forming subunit Orai1 (Vig et al., 2006; Park et al., 2009; Yuan et al., 2009) enables robust Ca²⁺ influx.

Although the molecular mechanisms of SOCE activation and the contributing proteins are well described (Liou et al., 2005; Roos et al., 2005; Mercer et al., 2006; Feske et al., 2006), little is known about regulatory mechanisms that modulate this important Ca²⁺ entry pathway. In view of the recent findings regarding endocannabinoids as modulators of the cellular ion homeostasis, this study was designed to elucidate the potential contribution of endocannabinoids to SOCE. For this purpose, AEA and its derivate N-arachidonoyl glycine (NAGly) (Bradshaw et al., 2009) which, in contrast to AEA, has no affinity to bind to CB₁ and CB₂ (Sheskin et al., 1997), were tested on SOCE activity. Our experiments revealed that NAGly, but not AEA, impairs the interaction of STIM1 with Orai1 and is highly effective in inhibiting SOCE, thus, demonstrating for the first time that the STIM1–Orai1-mediated SOCE machinery represents a specific target of an endocannabinoid. Although the metabolic origin of NAGly is not completely resolved, the wide distribution of this mediator in mammalian tissues (Huang et al., 2001) supports the notion that the inhibitory effect of NAGly on SOCE shown herein may also essentially contribute to the regulation of many physiological functions in vivo.

The effects of AEA and its derivate NAGly on SOCE signals in single cells of the human umbilical vein-derived endothelial cell line EA.hy926 were tested. Therefore, the classical Ca²⁺ re-addition protocol (Putney, 1986) was performed by stimulating fura-2/AM loaded cells with thapsigargin (Tg) and histamine in the absence of extracellular Ca²⁺ followed by addition of extracellular Ca²⁺ in the presence of the compound to be tested. While 10 µM AEA did not affect SOCE, it was significantly diminished in the presence of 10 µM NAGly (Fig. 1A), thus, indicating that the two structurally related endocannabinoids clearly differ in terms of their potency to impact SOCE in endothelial cells. Control experiments without cell stimulation ruled out the occurrence of Ca²⁺ entry without ER Ca²⁺ depletion in these cells (supplementary material Fig. S1A). In line with previous reports (Kohno et al., 2006), NAGly slowly mobilized Ca²⁺ from the ER in the absence of Tg/His, however, SOCE was not activated (supplementary material Fig. S1A). Moreover, once the Ca²⁺ entry maximally developed and reached a plateau upon Ca²⁺ addition, NAGly was able to reduce [Ca²⁺]_cyto (Fig. 1B). The impairment of Ca²⁺ entry by NAGly persisted only as long as the compound was present (Fig. 1C), indicating that the inhibitory effect of NAGly on SOCE is reversible.

Next, we tested the effect of NAGly on SOCE signals in endothelial cells that were stimulated exclusively with the inositol 1,4,5-trisphosphate-generating agonist histamine (His). Under these conditions [Ca²⁺]_cyto rapidly increased and reached a stable plateau of elevated [Ca²⁺]_cyto (Fig. 1D) which was shown to depend on continuous Ca²⁺ influx (Malli et al., 2003). The addition of NAGly at the plateau strongly attenuated [Ca²⁺]_cyto which recovered following NAGly removal (Fig. 1D). These findings demonstrated that NAGly also reversibly inhibits Ca²⁺ entry if cells are exclusively stimulated with an inositol 1,4,5-trisphosphate-generating agonist.

It was previously suggested that under such conditions of cell stimulation a receptor-activated Ca²⁺ entry (RACE) (Jousset et al., 2008) is activated in EA.hy926 cells in addition to SOCE, which is independent from ER Ca²⁺ depletion. Therefore, we tested whether or not the inhibitory effect of NAGly on Ca²⁺ entry remains the same if ER Ca²⁺ stores are depleted with Tg alone. These experiments showed that NAGly reduced SOCE signals to the same extent independently from the mode of ER Ca²⁺ depletion (supplementary material Fig. S1B). Moreover, our results indicate that cell stimulation with His is not essential for the inhibitory effect of NAGly on Ca²⁺ entry, pointing to SOCE rather than RACE as a likely target of this endocannabinoid in endothelial cells.

The inhibitory effect of NAGly on cytosolic SOCE signals in endothelial cells showed a clear concentration–response relationship for both the slopes (Fig. 1E) and amplitudes (supplementary material Fig. S1C) of respective rises in [Ca²⁺]_cyto upon Ca²⁺ addition. Notably, the half maximal inhibitory concentration (IC₅₀) of NAGly to attenuate the slope of rises in [Ca²⁺]_cyto upon Ca²⁺ addition in endothelial cells was found to be 4.22 (1.82–9.87) µM.

Analogous experiments were performed in RBL-2H3 cells, a mast cell model, which exhibits robust Ca²⁺ entry via SOCE (Schindl et al., 2002). In line with our findings in endothelial cells, NAGly was similarly effective to inhibit SOCE-driven [Ca²⁺]_cyto rises in RBL-2H3 cells (supplementary material Fig. S1D). The IC₅₀ of NAGly on SOCE perfectly matches the effective concentration of NAGly to reduce inflammatory processes (Burstein et al., 2011). Considering the fundamental role of SOCE in immune cells (Shaw and Feske, 2012), the inhibitory effect of NAGly on SOCE described herein may at least partially explain the immune-modulating actions of endocannabinoids. Moreover, SOCE is of utmost importance to stimulate the biosynthesis of multiple factors in endothelial cells (Graier et al., 1994), such as nitric oxide, prostaglandin (Lückhoff et al., 1988) and endothelin-1 (Brunner et al., 1994). Although a more detailed investigation is necessary, it is tempting to speculate that the effects of NAGly in the vasculature are, at least in part, due to its inhibitory potential on SOCE. Even though it has been estimated that the concentration of NAGly on potential sites of action vary between 5 and 24 µM (Wiles et al., 2006), the actual steady-state concentration of NAGly in vivo is not known, difficult to assess and depends on its rates of synthesis and hydrolysis. However, it is possible that local concentrations of lipid molecules, such as NAGly, might even exceed the µM range in vivo (Zaccagnino et al., 2009). Furthermore, we tested if a certain concentration of NAGly attenuates SOCE signals in a time-dependent manner. For this purpose, cells were pretreated with NAGly for different periods of time prior to Ca²⁺ addition after ER Ca²⁺ depletion. These experiments revealed that the magnitude of the inhibitory effect of NAGly on both the slopes (Fig. 1F) and amplitudes (Fig. 1F; supplementary material Fig. S1E) of SOCE correlated with the time period of preincubation. These findings indicate that NAGly might trigger a cellular signaling cascade in order to mediate the inhibitory effect on SOCE. Such an assumption is in line with a recent report demonstrating the activity of certain kinases, such as ERK1/2, to regulate SOCE activity (Pozo-Guisado et al., 2010). Since intact respiring mitochondria have been recognized to be fundamental for the activation, maintenance and termination of SOCE (Naghdi et al., 2010; Graier et al., 2007; Schwindling et al., 2010; Parekh, 2008), we tested the impact of NAGly on mitochondria in intact endothelial cells. In line with a previous study using isolated liver mitochondria (Zaccagnino et al., 2009), NAGly was indeed effective in acidifying and depolarizing mitochondria of endothelial cells (supplementary material Fig. S2), however, with a delay in time. Consequently, the temporal correlation (supplementary material Fig. S2) indicates that the NAGly-induced depolarization of mitochondria is not likely to be the cause of the instant repression of this Ca²⁺ entry pathway.

In order to verify whether the inhibitory effect of NAGly on SOCE is specific for this particular Ca²⁺ entry route, the impact of NAGly on Ca²⁺ influx through voltage-gated Ca²⁺ channels was tested using the pancreatic β-cell line INS-1 832/13 (INS-1). Like many other excitable cells, pancreatic β-cells are equipped with both L-type voltage-dependent Ca²⁺ channels (Nitert et al., 2008) and the SOCE machinery (Dyachok and Gylfe, 2001), which provides the possibility to test NAGly on these two distinct Ca²⁺ entry pathways in the same cell type. Treatment with 30 mM K⁺ in the presence of 20 mM glucose (Dyachok and Gylfe, 2001) triggered Ca²⁺ entry exclusively via voltage-gated Ca²⁺ channels in INS-1 cells, resulting in a fast elevation of [Ca²⁺]_cyto that was insensitive to NAGly (Fig. 2A). This is in line with a previous study using HEK cells, in which NAGly failed to inhibit the L-type voltage-dependent Ca²⁺ channel (Barbara et al., 2009). In agreement with our findings in endothelial and mast cells described above, NAGly also significantly reduced Ca²⁺ entry via SOCE in the pancreatic β-cell model (supplementary material Fig. S2B), emphasizing the general validity of the specific inhibitory effect of NAGly on SOCE described herein. Notably, the latter experiments were performed in the presence of the K⁺ channel opener diazoxide in order to prevent any Ca²⁺ entry through voltage-dependent Ca²⁺ channels (Dyachok and Gylfe, 2001).

To further investigate whether NAGly specifically targets the STIM1–Orai1-mediated SOCE, NAGly was tested on cells overexpressing both STIM1 and Orai1 (Mercer et al., 2006). As shown in Fig. 2C, NAGly was highly effective in attenuating increased SOCE signals, which led us to hypothesize that NAGly might reduce SOCE signals by disturbing the STIM1–Orai1 machinery itself.

The activation of SOCE upon ER Ca²⁺ depletion involves complex molecular processes starting with the oligomerization of STIM1 proteins, which subsequently redistribute to large STIM1 clusters in the subplasmalemmal area (Wu et al., 2006; Liou et al., 2007; Luik et al., 2008). In order to test whether or not NAGly affects STIM1 oligomerization, this process was monitored by measuring Förster resonance energy transfer (FRET) between the cyan fluorescent protein (CFP) and the yellow fluorescent protein (YFP) fused to the luminal N-terminus of STIM1 (Malli et al., 2008; Muik et al., 2008). Fast and strong ER Ca²⁺ depletion was achieved by cell treatment with Tg/His in the absence of extracellular Ca²⁺, which triggered a rapid increase of the FRET signal between CFP–STIM1 and YFP–STIM1. This elevated FRET signal remained stable in the absence of extracellular Ca²⁺, indicating a robust and constant oligomerization of STIM1 (Fig. 3A). Addition of NAGly did not reduce STIM1 oligomerization, indicating that this essential step in the activation of SOCE is insensitive to NAGly (Fig. 3A). Identical results were obtained in experiments where the fluorophores were at the C-terminus of STIM1, thus, being localized in the cytosol (supplementary material Fig. S3). In line with this finding, the stability of STIM1 clusters in ER Ca²⁺ depleted cells was not affected by NAGly either (Fig. 3B,C). We have previously reported that an increase in [Ca²⁺]_cyto reduces the amount of subplasmalemmal STIM1 clusters even in the virtual absence of ER Ca²⁺ refilling and despite almost unchanged oligomerization of STIM1 (Malli et al., 2008). The effect of [Ca²⁺]_cyto to destabilize STIM1 clusters was slightly delayed in the presence of NAGly, (Fig. 3C, left panel) that might reflect the reduced Ca²⁺ entry. However, the potency of [Ca²⁺]_cyto to reduce the appearance of subplasmalemmal STIM1 clusters remained unchanged (Fig. 3C, right panel).

Since NAGly was ineffective in impairing STIM1 oligomerization and clustering, we further focused on the interaction between STIM1 and Orai1 as a possible molecular target of NAGly. It was suggested that subplasmalemmal STIM1 clusters function as traps for Orai1 proteins (Xu et al., 2006), whereupon an intermolecular switching mechanism within the cytosolic domain of STIM1 allows electrostatic interaction with the C-terminal coiled-coil segment of Orai1 (Korzeniowski et al., 2010). Consistent with this intricate mechanism, the colocalization between YFP–STIM1 and CFP–Orai1 increased approximately twofold in response to ER Ca²⁺ depletion (Fig. 4A,C) as visualized by array confocal fluorescence microscopy setting the focal plane through the middle of individual cells. The addition of NAGly after ER Ca²⁺ depletion did not affect the increased colocalization between YFP–STIM1 and CFP–Orai1 in endothelial cells (Fig. 4B,C). Notably, these results were confirmed by confocal images across the surface of endothelial cells where a huge number of subplasmalemmal STIM1 clusters were detectable (supplementary material Fig. S4A,B).

Since colocalization experiments are not entirely conclusive on protein–protein interaction, a more sophisticated and detailed approach to study the interaction between STIM1 and Orai1 based on FRET was performed. In endothelial cells coexpressing Orai1–YFP and STIM1–CFP ER Ca²⁺ depletion decreased the donor CFP emission and increased the acceptor YFP emission simultaneously, resulting in a remarkable and long lasting increase in the FRET ratio (Fig. 4D), pointing to a stable interaction between the fusion proteins. Exposure to NAGly, however, led to a prompt reduction of the FRET signal due to a decrease in acceptor accompanied by an increase in donor emission intensities (Fig. 4D), indicating that the endocannabinoid diminished the molecular interaction between STIM1 and Orai1. In order to verify that NAGly indeed uncouples STIM1 and Orai1, we further examined their interaction by exchanging the donor and acceptor fluorescent proteins, respectively (i.e. measuring the FRET between STIM1–YFP and CFP–Orai1 (Fig. 4E; supplementary material Fig. S4C). In this setup, the basal ratio values (F_535[FRET]/F480_[DONOR]) were significantly lower, probably reflecting a different intracellular distribution of the two proteins. Nevertheless, ER Ca²⁺ depletion also evoked an increase in the STIM1–YFP/CFP–Orai1 FRET ratio, while NAGly treatment clearly decreased it (Fig. 4E; supplementary material Fig. S4C). These data demonstrate that NAGly efficiently diminishes STIM1–Orai1 FRET signals regardless of whether the FRET acceptor was fused to Orai1 or STIM1, thus, affirming that this endocannabinoid prevents STIM1–Orai1 interaction. Upon removal of NAGly the FRET signal between STIM1–CFP and Orai1–YFP recovered (Fig. 4F), which correlated with the reconstitution of Ca²⁺ entry via the SOCE pathway (Fig. 1C), thus, indicating that the inhibitory effect of NAGly on SOCE is a consequence of the diminished STIM1–Orai1 interaction. The inhibitory effect of NAGly on the interaction between STIM1 and Orai1 seemed to be specific, particularly in view of the finding that NAGly did not impact on STIM1 oligomerization (Fig. 3A; supplementary material Fig. S3). Moreover, in the respective control experiment NAGly did not affect the fluorescence ratio of Orai1–YFP and CFP–STIM1, which failed to show any increased FRET signal because of the different compartmentalization of the fluorophores (Fig. 4G). These findings indicate that the inhibitory effect of NAGly on the FRET signals between STIM1–CFP and Orai1–YFP or CFP–Orai1 and STIM1–YFP were indeed significant and not caused by an unspecific interference of NAGly with the fluorophores.

The putative involvement of mitochondria in the inhibitory action of NAGly on the STIM1–Orai1 interaction was tested using a mixture of antimycin A and oligomycin. This combination of mitochondrial toxins, which induces a strong and immediate depolarization of mitochondria, did not affect the STIM1–Orai1 interaction (supplementary material Fig. S4D). The subsequent addition of NAGly clearly reduced FRET signals between STIM1-CFP and Orai1–YFP in the presence of the mitochondrial toxins, supporting our assumption that the inhibitory effect of NAGly on SOCE is mainly independent of its effect on mitochondria.

In summary, our data provided herein unveiled a so far unknown inhibitory effect of the endocannabinoid NAGly on the signaling between STIM1 and Orai1. It is tempting to speculate that NAGly directly targets the STIM1–Orai1-containing SOCE machinery possibly by mimicking the negatively charged C-terminal coiled-coil segment of Orai1, which might decouple the positively charged cytosolic domain of STIM1 from Orai1 (Korzeniowski et al., 2010). Alternatively, NAGly might accomplish SOCE inhibition via its binding to certain receptors (Kohno et al., 2006), thereby activating a downstream signaling pathway which subsequently affects the interaction between STIM1 and Orai1. Notably, mRNA levels of both GPR18 and GPR55, two putative receptors of NAGly, were detected in the endothelial cell line used (supplementary material Fig. S5). Further experiments are necessary to clarify the exact molecular mechanism by which NAGly affects the SOCE machinery. However, our experiments revealed that NAGly, but not AEA, is highly effective in inhibiting SOCE specifically in different cell types by impairing the interaction of STIM1 with Orai1, a phenomenon that might serve as an explanation for its distinct physiological effects.

Cell culture materials were obtained from PAA laboratories (Pasching, Austria). N-Arachidonoylglycin was from Tocris Bioscience (Bristol, UK); anandamide and thapsigargin were purchased from Abcam® (London, UK); histamine, diazoxide, 1,4-dihydroxy-2,5-di-tert-butylbenzene (BHQ) and EGTA were from Sigma (Vienna, Austria).

Prior to experiments, cells were washed and maintained for 20 minutes in a HEPES-buffered solution composed of (in mM): 138 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 1 HEPES, 2.6 NaHCO₃, 0.44 KH₂PO₄, 0.34 Na₂HPO₄, 10 D-glucose, 0.1% vitamins, 0.2% essential amino acids and 1% penicillin/streptomycin; pH adjusted to 7.4 with NaOH. During the experiments cells were continuously perfused with a Ca²⁺-containing buffer, which consisted of (in mM) 145 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 10 D-glucose and 10 HEPES; pH adjusted to 7.4 with NaOH. In experiments where a Ca²⁺-free solution was applied, the CaCl₂ was replaced with 1 mM EGTA. INS cells were stimulated with a buffer containing (in mM): 113 NaCl, 30 KCl, 2 CaCl₂, 1 MgCl₂, 20 D-glucose and 10 HEPES; pH adjusted to 7.4 with KOH.

The EA.hy926 (human umbilical vein-derived endothelial) cell line, INS-1 832/13 pancreatic β-cells, and RBL-2H3 (rat basophil leukaemia) cells were used for this study and cultured as described previously (Alam et al., 2012; Malli et al., 2007; Schindl et al., 2002).

Cells were transfected with 1.5 µg of plasmid DNA (per 30 mm well) and 2.5 µg/well TransFast^TM transfection reagent (Promega, Madison, USA) in 1 ml of serum and antibiotic-free transfection medium 24 hours prior to experiments.

For the overexpression of YFP- or CFP-tagged STIM1 and Orai1 proteins, respectively, the following N- or C-terminally tagged constructs were used: CFP-Orai1, YFP-STIM1, CFP-STIM1 as well as Orai1-YFP, STIM1-YFP, and STIM1-CFP (Frischauf et al., 2009; Muik et al., 2008; Malli et al., 2008).

For cytosolic Ca²⁺ measurements single-cell Ca²⁺-imaging was performed as described previously (Graier et al., 1998; Malli et al., 2007). Briefly, cells were loaded with 2 µM Fura-2/AM (TEFLabs, Austin, TX, USA) for 45 minutes. Fura-2/AM-loaded cells were imaged using a fluorescence microscope described previously (Malli et al., 2007). For CFP/YFP-based FRET experiments cells were excited at 440 nm (440AF21; Omega Optical, Brattleboro, VT) and emission was gathered at 480 and 535 nm using an optical beam splitter (Dual View Micro-Imager™; Optical Insights, Visitron Systems) equipped with emission filters (480AF30, 535AF26; Omega Optical, Brattleboro, VT). Regions of interest covered whole individual cells. Results of FRET measurements are shown as the ratio of (F₅₃₅/F₄₈₀)/R₀ to correct for photobleaching and/or photochromism, as described recently (Waldeck-Weiermair et al., 2012; Malli et al., 2008).

High resolution imaging of YFP–STIM1 and CFP–Orai1 was performed with an array confocal laser scanning microscope (ACLSM), built on an inverse, fully automatic microscope (Axio Observer.Z1 from Zeiss, Göttingen, Germany) using a 100×/1.45 NA oil immersion objective (Plan-Fluor, Zeiss), and 405 nm (120 mW diode laser, Visitron Systems) and 488 nm (50 mW, VSLaserModul, Visitron Systems) laser light for exciting CFP and YFP, respectively. Emission was acquired with a CCD camera (CoolSNAP-HQ, Photometrics, Tucson, AZ, USA) using the emission filters ET480/40m for CFP and ET535/30m for YFP (Chroma Technology Corporation, VT, USA). All devices were controlled by VisiView Premier acquisition software (Visitron Systems). Image analysis was performed with MetaMorph 7.7.0.0 (Visitron Systems) using the integrated morphometric analysis tool and the colocalization application.

EA.hy926 cells were loaded with 150 nM tetramethylrhodamine methyl ester (TMRM) for 30 min at room temperature. All subsequent experiments were performed in quench mode with solutions as indicated in Chemicals and buffer solutions but containing 150 nM TMRM. TMRM fluorescence intensities of single cells were measured over non-mitochondrial (nuclear) regions using a 40× oil immersion objective on a Zeiss AxioVert inverted microscope (Zeiss, Austria) equipped with a polychromator illumination system (Visitron Systems, Germany) and a CCD camera (Photometrics, AZ, USA). TMRM was excited at 548 nm and emission was detected at 570 nm.

EA.hy926 cells stably expressing mitochondrial targeted pericam were used to measure the mitochondrial pH. Cells were illuminated at 480 nm and emission was collected at 535 nm using the same imaging system described above. At the end of each individual experiment 2 μM FCCP was used to induce maximal mitochondrial acidification.

RNA was isolated from Ea.hy926 cells using a Total RNA isolation kit (PEQLAB Biotechnologie GmbH, Erlangen, Germany) and it was reverse transcribed using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, USA). The analysis of the expression of the target genes was performed by conventional polymerase chain reaction (PCR) using GoTaq Green master mix (Promega, Madison, WI, USA) and real-time PCR using QuantiFast SYBR Green RT-PCR kit (Qiagen, Hilden, Germany) on LightCycler 480 (Roche Diagnostics, Vienna, Austria). RNA polymerase II (RPOL2) was used as a housekeeping control. Primers for RPOL2, GPR18 and GPR55 were obtained from Invitrogen (Vienna, Austria) and their sequences (5′–3′) are as follows: RPOL2 for: CATTGACTTGCGTTTCCACC, RPOL2 rev: ACATTTTGTGCAGAGTTGGC; GPR18 for: CCACCTTCCTCATGAACCTC, GPR18 rev: GACCGTAGACTACCAGATCG; GPR55 for: TTGACGGTGTCAACGAGCTG, GPR55 rev: TAAGGAAGGTGCTGAAGCCA.

Data shown are means ± s.e.m., and n indicates the number of cells/independent experiments. Statistical analyses were performed with paired and unpaired Student's t-tests, and P<0.05 was considered to be significant.

We thank Therese Macher and Dr Rene Rost for their excellent technical assistance, Dr C. J. S. Edgell (University of North Carolina, Chapel Hill, NC, USA) for the EA.hy926 cells and Dr C. B. Newgard (Duke University, Durham NC, USA) for the INS-1 cells. Moreover we thank Prof. Christoph Romanin (University of Linz, Austria) for sending us the FP-tagged STIM1 and Orai1 constructs. We are grateful to Dr Maud Frieden (University of Geneva), Prof. Klaus Groschner, Mag. Karin Osibow, and Dr Michael Poteser (Medical University of Graz, Austria) for their critical comments.