The SOCE (store-operated Ca2+ entry) pathway is a central component of cell signalling that links the Ca2+-filling state of the ER (endoplasmic reticulum) to the activation of Ca2+-permeable channels at the PM (plasma membrane). SOCE channels maintain a high free Ca2+ concentration within the ER lumen required for the proper processing and folding of proteins, and fuel the long-term cellular Ca2+ signals that drive gene expression in immune cells. SOCE is initiated by the oligomerization on the membrane of the ER of STIMs (stromal interaction molecules) whose luminal EF-hand domain switches from globular to an extended conformation as soon as the free Ca2+ concentration within the ER lumen ([Ca2+]ER) decreases below basal levels of ~500 μM. The conformational changes induced by the unbinding of Ca2+ from the STIM1 luminal domain promote the formation of higher-order STIM1 oligomers that move towards the PM and exposes activating domains in STIM1 cytosolic tail that bind to Ca2+ channels of the Orai family at the PM and induce their activation. Both SOCE and STIM1 oligomerization are reversible events, but whether restoring normal [Ca2+]ER levels is sufficient to initiate the deoligomerization of STIM1 and to control the termination of SOCE is not known. The translocation of STIM1 towards the PM involves the formation of specialized compartments derived from the ER that we have characterized at the ultrastructural level and termed the pre-cortical ER, the cortical ER and the thin cortical ER. Pre-cortical ER structures are thin ER tubules enriched in STIM1 extending along microtubules and located deep inside cells. The cortical ER is located in the cell periphery in very close proximity (8–11 nm) to the plasma membrane. The thin cortical ER consists of thinner sections of the cortical ER enriched in STIM1 and devoid of chaperones that appear to be specialized ER compartments dedicated to Ca2+ signalling.
Cytosolic Ca2+ elevations control a large array of cellular functions, notably the contraction of cardiac and skeletal muscle cells, the secretion of hormones and neurotransmitters by endocrine cells and neurons, and the activation of gene transcription in haemopoietic cells. Activation of PLC (phospholipase C)-coupled membrane receptors results in the generation of the intracellular messenger inositol triphosphate that releases Ca2+ from the ER (endoplasmic reticulum), the major internal Ca2+ store of cells. Owing to the limited capacity of internal Ca2+ stores, the release of Ca2+ from ER is linked to the influx of Ca2+ across PM (plasma membrane) channels, whose opening sustains long-term Ca2+ signals and promotes the refilling of the depleted stores. The coupling between ER Ca2+ depletion and Ca2+ entry is known as SOCE (store-operated Ca2+ entry), formerly termed capacitative Ca2+ entry [1,2]. SOCE is a universal mechanism essential for the maintenance of physiological Ca2+ levels within the ER lumen permissive for protein folding, and for the generation of long-term Ca2+ signals that drive gene expression in several cell types including immune cells . The most studied SOCE channel is the CRAC (Ca2+-release-activated Ca2+) channel characterized by a high Ca2+ selectivity and tiny conductance . Although the concept of SOCE was formalized 20 years ago , the molecular players were unravelled only recently by genome-wide RNAi (RNA interference) screens in human and Drosophila S2 cells, the two main components being the ER Ca2+-sensing STIMs (stromal interaction molecules) [5–7] and the PM-located channel pore forming subunits Orai [8–10]. Co-overexpression of these two molecules leads to a large increase in SOCE influx in all of the cellular systems tested so far [11,12].
Opening the gate: molecular mechanisms of SOCE activation
The single-pass ER transmembrane protein STIM1 and its homologue STIM2 has dual functions as it both senses the Ca2+-filling sate of intracellular ER Ca2+ stores and activates PM SOCE channels. STIM1 senses the free Ca2+ concentration within the ER lumen via its N-terminal Ca2+-binding EF-hand motif that protrudes inside the ER lumen (STIM1 Kd ~0.2–0.6 mM ). At physiological resting [Ca2+]ER of 400–600 μM , STIM1 is in the Ca2+ -bound state and adopts a globular conformation that promotes the formation of stable dimers via interactions between the proximal coiled-coil domains of its cytosolic tail . Upon store depletion, Ca2+ unbinding from STIM1 EF-hand domain fused to a SAM (sterile α-motif) domain induces a conformational change that leads to the formation of higher-order STIM oligomers and exposes activating domains in STIM1 cytosolic tail that bind to and activate Orai channels at the plasma membrane  (Figure 1). This sequence of 100 amino acids known as CAD (for CRAC-activation domain) or SOAR (for STIM1–Orai-activating region) contains the second of the two successive coiled-coil domains of STIM1 C-terminal tail and activates the Orai1 channel via electrostatic interactions with its C-terminal coiled-coil domain [17–21]. Both the luminal EF–SAM domains and the CAD/SOAR domain on the STIM1 cytosolic tail are critical for SOCE signalling, because exogenous expression of STIM EF-hand mutants that do not bind Ca2+ leads to constitutively active SOCE [7,22], as does expression of the CAD/SOAR domain or the entire cytosolic tail of STIM1 [23–25]. STIM2 has 60% similarity in sequence, and its overexpression was initially shown to reduce SOCE , but this isoform responds to smaller changes in Ca2+ than does STIM1 and is now considered to regulate basal cytosolic and ER Ca2+ levels .
Molecular assembly of SOCE complexes
In cells with replete stores, STIM1 is highly dynamic and forms comet tails structures on live fluorescence microscopy images, a phenomenon that involves the binding of STIM1 to the microtubule plus-end-tracking protein EB1 (end-binding protein 1) via a SXIP (Ser-Xaa-Ile-Pro) motif on the STIM1 cytosolic tail [28,29]. This movement facilitates the extension of the ER towards the periphery of the cell through the TAC (tip-attachment complex) mechanism, a process whereby ER tubules extend by attaching to the growing plus ends of microtubules . The Ca2+ depletion of the ER abolishes the interaction between STIM1 and EB1 and promotes STIM1 oligomerization, a step that is sufficient to induce the translocation of STIM1 towards the plasma membrane, where the oligomers accumulate in membrane-associated punctate structures that overlap with Orai1 membrane clusters and with Ca2+-entry sites [31–33]. This process is ATP-independent  and is promoted by the interaction between STIM1 C-terminal polybasic tail and membrane phospholipids . The close proximity between the ER and the PM at Ca2+-entry sites enables productive interactions between STIM1 CAD/SOAR and the Orai1 C-terminal region, and possibly between STIM1 distal C-terminal lysine-rich domain and a proline-rich region on the N-terminus of Orai1 . FRET (fluorescent resonance energy transfer) experiments revealed that the formation of STIM1 oligomers precedes the interaction between STIM1 and Orai1 , and enforced oligomerization of STIM1 by the addition of rapamycin-binding domains to its luminal tail is sufficient to initiate SOCE, indicating that the oligomerization of STIM1 is a rate-limiting step in the activation of SOCE.
Morphological changes associated with SOCE activation
Interestingly, the location of the STIM1–Orai1 contact sites at the cell periphery appears to be predetermined, because, in response to ER Ca2+ depletion, CFP (cyan fluorescent protein)-tagged STIM1 redistributes to pre-formed puncta containing the constitutively active YFP (yellow fluorescent protein)–STIM1 EF-hand mutant , and repeated store depletion and refilling with agonists induces the apparition of STIM1 clusters at the same locations during sequential stimulations . Under the electron microscope, the accumulation of ER-derived compartments near the plasma membrane can be seen to form characteristic ER structures that we termed cortical ER because of their location at the cortex of cells. Both store depletion and STIM1 overexpression enhanced the number and the length of the cortical ER, although no obvious rearrangement of the bulk ER could be observed with ER-targeted fluorescent dyes [37,38]. The cortical ER is continuous with deeper regions of the ER, is deprived of ribosomes on the site facing PM, and is often decorated by microtubules on its cytosolic side. The cortical ER comprises thinner segments (thickness 24 ± 0.4 nm compared with 73 ± 3 nm for the bulk cortical ER) that lack ribosomes also on the cytosolic side. The thin cortical ER is enriched in STIM1 and is devoid of proteins containing the ER-targeting KDEL (Lys-Asp-Glu-Leu) motif such as BiP (immunoglobulin heavy-chain-binding protein), suggesting that this specialized compartment derived from the ER might be dedicated to cell signalling. Another ER structure exhibits the same morphological properties as the cortical ER, but is located deep inside cells, and we termed this compartment pre-cortical ER because it is devoid of contacts with the PM . The pre-cortical ER is frequently found extending along microtubules, implying a possible role of microtubule in the formation of the cortical ER . A yeast ER transmembrane protein resembling STIM1, Ist2, also induces the formation of cortical ER upon dimerization, and Ist2-mediated cortical ER formation was shown to require microtubules and EB1 . Microtubules are not required for SOCE activation, however, because microtubule depolymerization with taxol or nocozadole does not abrogate SOCE and only disturbs the movement of STIM1 comets occurring before the Ca2+ depletion of the ER [28,40]. The mechanism of STIM1-induced formation of the cortical ER induced by store depletion thus appear to be more complex than the mechanism of Ist2-induced cortical ER formation in yeast, and might involve additional cellular elements. Like Ist-2, STIM proteins possess a polybasic cytosolic tail that is thought to mediate their plasma membrane targeting by interacting with membrane lipids. Accordingly, the C-termini of both STIM1 and STIM2 bind to PtdIns(4,5)P2 on purified liposomes , whereas in cells, PI4K (phosphoinositide 4-kinase) inhibition by wortmannin reduces SOCE current  and the depletion of multiple membrane phosphoinositides, but not of PtdIns(4,5)P2, PtdIns4P or PtdIns(3,4,5)P3 alone, reduces STIM puncta formation. PtdIns(4,5)P2 depletion has minimal effects on pre-assembled STIM1 puncta induced by prior store depletion, however, and the inhibitory effects of phosphoinositide depletion on STIM1 membrane clustering can be rescued by the overexpression of Orai1 . The lysine-rich C-terminal domain is thus dispensable for STIM1 membrane targeting, which depends more stringently on the interactions between STIM1 and Orai channels, and phosphoinositides appear to play a facilitating role during the SOCE-activating phase by enhancing the recruitment of STIM1 to the PM.
STIM proteins were initially described as adhesion proteins involved in intercellular interactions , and biotinylation assays indicate that during store depletion, a fraction of STIM1 can reach the cell surface . Another study confirmed the presence of externally exposed STIM1, but not of STIM2, and did not observe any increase in membrane expression following store depletion . However, neither fluorescent protein-tagged STIM1 nor the constitutively active EF-hand mutant can be detected at the cell surface [6,44]. The larger size of the fusion proteins or the multimeric state of the constitutively active mutant might prevent membrane surface exposure of the exogenously expressed STIM proteins. The PM STIM1, whose EF-hand domain faces the extracellular space, has been proposed to be part of the SOCE complex and to play a regulatory role in Ca2+ influx. Antibodies against the N-terminus of STIM1 applied to the extracellular medium have been shown to inhibit the CRAC current, but the interpretation of these results is complicated by the inhibitory effects of these antibodies on the Ca2+ influx resulting from the expression of the constitutively active EF-hand mutant, which, as discussed above, is not exposed to the extracellular side [22,44].
Studies using chemical inducible linkers of various lengths to anchor the ER to the PM and manipulate the distance between the two membranes suggested that the gap between STIM1 and Orai1 at Ca2+-entry sites is approximately 11–14 nm , i.e. too wide to allow direct interactions between the two protein partners. Our subsequent EM measurements indicated that the gap is in fact smaller, averaging 8.3 nm in Epon sections and 11.3 nm in cryosections of HeLa cells expressing exogenous STIM1 and treated with thapsigargin . This suggests that direct interactions between STIM1 and Orai can indeed take place at the junction between cortical ER and the PM, consistent with in vitro studies showing that insertion of recombinant STIM1 and Orai1 in purified membrane vesicles is sufficient to generate Ca2+ influx . Although overwhelming evidence indicates that STIM1 interacts directly with Orai1 to control the channel gating, other proteins have been shown to be co-recruited to the Ca2+-entry sites and to participate in the SOCE process. One expected player is the SERCA (sarcoplasmic/endoplasmic reticulum Ca2+-ATPase) that pumps Ca2+ ions into the ER lumen to refill the depleted stores. Store depletion increased the extent of co-localization between SERCA2 and STIM1 on immunofluorescence images [47,48] and enhanced the amount of SERCA3 that could be retrieved by co-immunoprecipitation of STIM1, suggesting that the two proteins interact in active SOCE complexes  (Figure 1). Consistent with a STIM1–SERCA interaction, overexpression of SERCA2a disturbed the interaction between STIM1 and Orai1 in human smooth muscle cells . Functional data also suggest that active SERCA are located close to Ca2+-entry sites and sequester the entering Ca2+ in the ER, thereby preventing potentially toxic global cytosolic Ca2+ elevation during store refilling .
Interestingly, quantitative electron microscopy revealed that a different amount of cortical ER was recruited at the apical and basolateral membranes of HeLa cells, indicating that SOCE molecules have particular tropisms for specific membranes . The recruitment of STIM1 and Orai clusters at specific locations of the plasma membrane provides a morphological substrate for the generation of localized cytosolic Ca2+ signals. In the immune system, STIM1 and Orai1 were shown to translocate to the immunological synapse formed at the interface between T-cells and antigen-presenting cells, and to maintain the localized sustained Ca2+ elevation required for the activation of T-cells . The dominant-negative Orai1 mutant was also recruited to the immunological synapse, indicating that the generation of local Ca2+ signals is not required to accumulate or retain SOCE molecules at this cellular site. In pancreatic acinar cells, STIM1 accumulates at the basolateral pole upon ER depletion, whereas the Ins(1,4,5)P3 ER Ca2+-release channel receptor is predominantly located on the apical pole . The basolateral ER regions enriched in STIM1 have a morphology typical of the cortical ER by electron microscopy and are devoid of attached ribosomes. Surprisingly, a fraction of the STIM1 partner Orai1 was found to co-localize with the Ins(1,4,5)P3 receptor at the apical pole, irrespective of store depletion . The preferential recruitment of STIM1 to the basolateral pole thus does not appear to reflect a preferential distribution of its target Orai1 channel in this particular membrane. Instead, the lipid composition of the membrane appears to play a facilitating role in the recruitment of cortical ER structures. Lipid rafts are special PM domains that favour protein interaction and the assembly of signalling protein complexes. Store depletion promoted the recruitment of STIM1 to ER structures facing lipid raft domains enriched in the TRPC (canonical transient receptor potential) 1 cation channel, and both SOCE and STIM1 puncta formation were prevented by the disruption of lipid rafts with methyl-β-cyclodextrin or Filipin-III . The involvement of TRPC channels in SOCE is still controversial, with several groups reporting that TRPC channels interact with STIM1 and/or with Orai channels and therefore contribute to SOCE influx [56,57], and others claiming that TRPC channels function independently of STIM and Orai . Membrane lipid rafts may act as a switching platform that convert TRPC channels from receptor-operated channels into store-operated channels by favouring their interaction with cortical ER structures enriched in STIM1 and facing the lipid rafts .
Closing the gate: termination of the SOCE process
The opening of Ca2+-entry channels is tightly regulated, because aberrant Ca2+ elevations can lead to inappropriate cellular responses and can damage cells. SOCE channels self-inactivate as the concentration of Ca2+ ions on the cytosolic side increases, a process known as CDI (Ca2+-dependent inactivation). CDI was characterized early on by patch-clamp recordings and shown to involve both a fast and a slow component, with the fast CDI requiring localized Ca2+ elevations very close to the channel mouth . Mutagenesis experiments indicated that the sites of Orai1 crucial for fast CDI overlap with the sites important for ion permeation as substitution of acidic amino acids lining the pore of the Orai1 channel caused a loss in Ca2+ selectivity and diminished the fast CDI of the currents . Further analysis of STIM1 truncation mutants indicated that the CAD/SOAR domain induces currents that lack fast inactivation and revealed that a short sequence located distally to CAD/SOAR on STIM1 C-terminal domain is essential for the fast inactivation. This CMD (CRAC-modulatory domain) (amino acids 474–485) contains seven negatively charged amino acids whose neutralization significantly reduced the fast inactivation of co-expressed Orai1 [62–64]. The rapid CDI was shown further to involve the Ca2+-dependent binding of CaM (calmodulin) to a membrane-proximal N-terminal domain of Orai1, and mutations that eliminate CaM binding abrogated CDI . Interestingly, CaM also binds to the polybasic tail of STIM1 and STIM2 in a Ca2+-dependent manner , suggesting that the local cytosolic Ca2+ elevations are sensed by CaM, which then binds to the STIM1 C-terminus and/or to the N-terminus of Orai1 to induce the rapid inactivation of SOCE channels. Mitochondria were proposed to minimize the slow Ca2+-dependent inactivation by sequestering the Ca2+ ions entering across SOCE channels [51,66,67]. However, given the proximity between the cortical ER and the PM, the notion that mitochondria can have an impact on the local Ca2+ concentration at the channel mouth has been challenged, and measurements with Ca2+-sensitive probes anchored on the outer mitochondrial membrane revealed that subplasmalemmal mitochondria are not exposed to high Ca2+ microdomains . Mitochondria might have pleiotropic effects on the SOCE process, as the impairment of the mitochondria buffering capacity was shown to impair SOCE activation in endothelial cells [69,70].
STIM1 oligomerization and translocation are both reversible events [32,36], but the mechanisms that control the deoligomerization of STIM1 and its retrieval from the PM have been comparatively less studied than the mechanisms that control the recruitment of STIM1 to the PM during SOCE activation. ER store refilling correlates temporally with SOCE channel inactivation and with the retrieval of STIM1 to deeper ER regions. By analogy with the mechanism that controls STIM1 oligomerization, it is assumed that the deoligomerization of STIM1 is controlled by the binding of Ca2+ to the STIM1 luminal domain. Accordingly, reversible changes in the STIM1 oligomerization state can be observed in cells stimulated with physiological Ca2+-mobilizing agonists that induce oscillatory changes in [Ca2+]ER (Figure 2). This suggests that an increase in [Ca2+]ER is sufficient to induce the deoligomerization of STIM1 during the early phase of SOCE activation. However, the STIM1 oligomers that accumulate in the cortical ER are likely to be stabilized further by the electrostatic interactions occurring between the STIM1 CAD/SOAR domain and Orai channels and between the STIM1 lysine-rich tail and membrane lipids. Whether the conformational changes induced by the transition of the luminal domain from the Ca2+-unbound to the Ca2+-bound state are sufficient to disrupt the ER–PM interactions is not known, but cytosolic Ca2+ elevations might play a role in the termination of the SOCE process. A cytosolic protein with two Ca2+-sensitive EF-hand motifs, CRACR2 (CRAC regulator 2), was shown recently to contribute to the stabilization of STIM–Orai complexes at low cytosolic Ca2+ levels, suggesting that localized cytosolic Ca2+ elevations might promote the termination of SOCE by inducing the dissociation of CRACR2 from the assembled STIM1–Orai1 complex .
Reversibility of the [Ca2+]ER-induced changes in STIM1 oligomerization
SOCE is a fundamental cellular event whose activation and inactivation is finely regulated to avoid inappropriate Ca2+ elevations that can be deleterious for cells. Since the discovery of the STIM and Orai family of proteins, much knowledge has been gained on the molecular mechanisms of SOCE activation and on the accompanying morphological changes leading to the formation of cortical ER compartments dedicated to Ca2+ signalling. The next challenge will be to understand the mechanisms that control the dissociation of STIM1–Orai1 complexes at the ER–PM interface and the retrieval of the cortical ER during the resolving phase of SOCE.
Signalling 2011: a Biochemical Society Centenary Celebration: A Biochemical Society Focused Meeting held at the University of Edinburgh, U.K., 8–10 June 2011. Organized and Edited by Nicholas Brindle (Leicester, U.K.), Simon Cook (The Babraham Institute, U.K.), Jeff McIlhinney (Oxford, U.K.), Simon Morley (University of Sussex, U.K.), Sandip Patel (University College London, U.K.), Susan Pyne (University of Strathclyde, U.K.), Colin Taylor (Cambridge, U.K.), Alan Wallace (AstraZeneca, U.K.) and Stephen Yarwood (Glasgow, U.K.).
cyan fluorescent protein
CRAC regulator 2
end-binding protein 1
fluorescent resonance energy transfer
sarcoplasmic/endoplasmic reticulum Ca2+-ATPase
store-operated Ca2+ entry
stromal interaction molecule
canonical transient receptor potential
yellow fluorescent protein
We thank Dr Maud Frieden for a critical reading of the paper before submission.
Our work is supported by the Swiss National Science Foundation [grant number 3100A0-118393].