The importance of SOCE in cell physiology, and its likely contribution to the pathology of several diseases were recognized two decades ago (for review see ). However, not until the identification of the molecular components of this pathway has a precise and in depth study of the actual role of SOCE been possible. In human beings, a nonsense mutation that results in STIM1 depletion has been linked to immunodeficiency, autoimmune haemolytic anaemia, thrombocytopenia, muscular hypotonia and abnormal enamel dentititon , a clinical phenotype that is similar to that caused by mutations in ORAI1 . In mice, STIM1 deficiency is perinatally lethal, with the majority of STIM1-deficient mice created by homologous recombination displaying severe growth retardation and dying shortly before, or just after birth [50, 61] and those created by gene-trap insertion dying around 4–6 weeks following birth . STIM2-deficient mice display a less severe growth retardation phenotype, and begin to die spontaneously from around 4–8 weeks after birth depending on the genetic background, with few survivors beyond 30 weeks [50, 63]. Human and mouse cells isolated from STIM-deficient individuals as well as genetically manipulated cell lines have provided important information regarding the mechanisms by which STIM proteins might regulate cell physiology and result in pathologies exhibited by STIM deficiency.
The importance of SOCE in immune function is underpinned by the identification of at least four families of patients with severe combined immunodeficiency (SCID) associated with a severe reduction in SOCE in T lymphocytes, B lymphocytes and fibroblasts [60, 64–66]. In the two families that have been analysed, immunodeficiency is caused by either homozygous mutations in the pore forming region of ORAI1 or a homozygous nonsense mutation in STIM1, both of which result in abrogation of SOCE and CRAC currents [14, 60]. The STIM1 nonsense mutation results in a frameshift and premature termination, which would result in a truncated STIM1 protein containing only the N-terminal EF-hand domain and a small segment of the adjacent SAM domain . However, the predicted truncated protein was not detected in patient fibroblast cells, and strongly reduced transcript levels of STIM1 suggested that the mutation likely results in nonsense-mediated mRNA decay and a STIM1 null phenotype.
The SCID phenotype in STIM1-deficient patients is characterized by life-threatening viral, bacterial and fungal infections from an early age. Total lymphocyte counts were within the normal range in these patients, and the percentages of CD4+ and CD8+ T cells and of CD19+ B cell subsets were also normal in all three patients . In contrast, a lack of antigen-specific antibody responses and impairment in T-cell activation is evident by a marked reduction in proliferation of CD4+ T cells in vitro. Together these data indicate that the immunodeficiency in STIM1-deficient patients is likely not due to aberrant lymphocyte development, but rather is caused mainly by impaired lymphocyte function.
Analysis of immune cells in STIM1-deficient mice bearing a T-cell specific deletion in CD4+ T cells demonstrated normal lymphocyte development despite undetectable SOCE and CRAC channel currents in naïve and stimulated T cells . In common with human STIM1-deficient patients, T-cell activation was severely compromised in STIM1-deficient CD4+ T cells, as assessed by a failure to produce interleukin (IL)-2, interferon (IFN)-γ and IL-4 in response to T-cell receptor stimulation. The expression of these critical immunomodulatory cytokines are under complex regulation, and their promoters are rich in binding sites for Ca2+-sensitive transcription factors such as NF-κβ, NFAT and CREB which act in a combinatorial fashion to ensure optimal transcriptional output . In SCID patients bearing mutations in ORAI1, nuclear extracts from T cells contain virtually no NFAT binding activity, implicating this transcription factor as a key mediator of the disease . In resting T cells, NFAT normally resides in the cytoplasm in an inactive phosphorylated form, and T-cell receptor activation induces SOCE and activation of the serine/threonine phosphatase calcineurin . Calcineurin dephosphorylates NFAT and mediates its translocation into the nucleus where it exerts transcriptional regulation on the IL-2 promoter . Calcineurin-mediated nuclear import is balanced by nuclear export, involving nuclear kinases such as GSK3β, casein kinase I and DYRK1A that rephosphorylate NFAT . In stimulated STIM1-deficient CD4+ T cells, nuclear translocation of NFAT is severely reduced when compared to control T cells, resulting in diminished cytokine production .
In contrast to STIM1-deficient CD4+ T cells, SOCE and ICRAC in resting STIM2-deficient CD4+ T cells is virtually normal, and only slightly affected in activated CD4+ T cells, where STIM2 expression is normally up-regulated . Despite this, T-cell activation-induced production of IL-2 and IFN-γ is reduced, correlating with reduced NFAT nuclear residency. However, in stark contrast to STIM1-deficient CD4+ T cells, the ability to drive NFAT to the nucleus is hardly affected in STIM2-deficient CD4+ T cells, but these cells are completely unable to retain NFAT nuclear localization. Together, these results clearly suggest different, but synergistic, roles for STIM1 and STIM2 in the regulation of NFAT transcriptional activity: STIM1 as a driver of NFAT nuclear localization, and STIM2 as an inhibitor of NFAT nuclear export. Both activities appear necessary for optimal NFAT transcriptional activity in CD4+ T cells.
In support of the proposition that STIM1 and STIM2 synergize in regulating lymphocyte function, mice double deficient for STIM1 and STIM2 in CD4+ T cells develop a lymphoproliferative disorder characterized by splenomegaly, lymphadenophathy, dermatitis and blepharitis, or inflammation of the eyelid margins, and infiltration of leucocytes into many organs . Whereas all T- and B-lymphocyte subsets are normal in single STIM-deficient mice, double STIM1/STIM2-deficiency results in a 90% reduction in CD4+CD25+FoxP3 regulatory T cells (Tregs). The resulting lymphoproliferative disorder is ameliorated by supplementation of STIM1-/STIM2-deficient mice with wild-type Tregs .
Tregs function to dampen immune responses to self-antigens. Naturally occurring Tregs are thought to develop from CD4+ CD8+ T cells, mediated by FoxP3, a member of the forkhead family of transcription factors . Notably, binding sites for NFAT2 reside in the FOXP3 promoter and can positively regulate FOXP3 expression in vitro. Furthermore, a complex consisting of NFAT and FoxP3 proteins is recognized to bind to the IL-2 promoter . It appears to be the combined STIM1-/STIM2-mediated reduction in activity of NFAT in developing double-deficient Treg cells that result in the vastly reduced number of Treg cells in double-deficient mice [17, 50, 74]. However, this is unlikely to be the only mechanism involved, because complete abrogation of NFAT activity in T cells of STIM1-deficient mice does not affect Treg numbers  and mice that are double deficient in NFAT1 and NFAT4, or NFAT1 and NFAT2 also have normal Treg numbers . Moreover, several studies indicate that it is in fact a low level of NFAT activity that is required for Treg function because CsA treatment in patients can increase Treg-like function in some instances, and FoxP3, whilst interacting with NFAT at the IL-2 promoter, acts as a repressor of NFAT-mediated IL-2 transcription . Thus it seems plausible that combined STIM1/STIM2 deficiency impinges not only on the NFAT signalling pathway, but also on additional signalling pathways critical for mediating the development or homeostasis of Treg cells.
The function, but not the development, of other immune cell subsets is also impaired in STIM1-deficient mice, in addition to T lymphocytes, including mast cell degranulation, macrophage-mediated phagocytosis and platelet activation (for review see ). The other clinical symptoms of patients bearing mutations in STIM1 and ORAI1, however, suggest that these proteins are important for the development of other tissues, most notably skeletal and smooth muscle, and neural tissue .
Several studies now point to a critical role for STIM1 in controlling the Ca2+-mediated events leading to correct muscle development and function. Patients with STIM1 deficiency suffer from congenital non-progressive muscle weakness  and STIM1-deficient mice bearing a gene trap in the C-terminus of STIM1 have reduced skeletal muscle cross sectional area concomitant with a marked increase in inter-fibre connective tissue . A number of the fibres have centrally, rather than peripherally, placed nuclei, and a great increase in large swollen mitochondria between myofibres and in the subsarcolemma. These features are considered hallmarks of centronuclear myopathy, a subtype of congenital myopathies characterized by weak muscles, abundant mitochondria and an increase in fat and connective tissue . Mutations in several genes encoding proteins involved in cytoskeletal remodelling and endocytosis, including myotubularin, dynamin and amphiphysin, have been identified as the cause of these centronuclear myopathies . Individual myotubes prepared from STIM1-deficient mice display essentially no SOC entry, and the lack of myosin heavy chain expression indicates defective muscle development . Indeed, siRNA-mediated knockdown of STIM1 prior to the onset of differentiation in human myoblasts significantly reduces differentiation capacity in a dose-dependent manner . Effects on muscle differentiation require an early reduction of SOCE during development, because inhibition of SOCE at time-points after the onset of differentiation fails to influence myotube formation. Thus the amplitude of SOCE critically determines a differentiation signal that drives the early differentiation of myoblasts, potentially at a stage prior to hyperpolarization of the plasma membrane to –70 mV . Hyperpolarization occurs by dephosphorylation-mediated up-regulation of Kir2.1 K+ channel activity, which is maximal 6 hrs after induction of differentiation . Of note, inhibition of myoblast fusion decreases Kir2.1 K+ channel-mediated hyperpolarization, and STIM1-deficient myoblast cells fail to cluster their nuclei, a hallmark of human myoblast fusion. Moreover, STIM1 overexpression accelerates differentiation in human myoblasts  and in C2C12 mouse myoblasts (Fig. 2; S. Graham, L. Johnstone, unpublished results), likely by increasing fusion events. A role for the Ca2+ activated calcineurin/NFAT pathway in the regulation of myoblast fusion has been demonstrated by the diminished muscle mass in NFAT2-deficient mice  and the abrogation of myoblast fusion in vitro by inhibitors of calmodulin activity . Moreover, increasing intracellular Ca2+ induces myoblast fusion and buffering intracellular Ca2+ levels inhibits myoblast fusion . Together these observations suggest that STIM1-mediated Ca2+ dependent signalling processes are likely to regulate processes necessary for myoblast fusion.
Figure 2. Immunofluorescent staining of myosin heavy chain (MHC) in fixed C2C12 cells overexpressing STIM1. Cells transiently overexpressing STIM1 or empty vector were fixed at day 6 after induction of differentiation and immunostained for MHC (green) and DAPI (blue) to identify myotubes (mt). The experiment was performed in duplicate. The differentiation index (percentage of nuclei associated with MHC) was calculated by analysing 16 fields of view per group. Vector – 40.2%, STIM1 – 59.7% (P= 0.009).
Download figure to PowerPoint
Although STIM1 expression is necessary at the early stages of muscle development, up-regulation of STIM1 expression at subsequent stages of myoblast differentiation in vitro and in vivo indicate that STIM1 has additional roles in mature muscle function [47, 48, 83]. It has long been appreciated that skeletal muscle contraction and relaxation are controlled by the precise release and re-uptake of Ca2+ from the specialized sarco/endoplasmic reticulum (SR) Ca2+ store within skeletal muscle fibres. These highly specialized cells are designed with specific tubular invaginations of the plasma membrane, T tubules, which penetrate deep into the cytoplasm. Terminal portions of the SR flank each T tubule, and are physically coupled to it by interactions between L-type calcium channels located on the T tubule membrane and Ryanodine receptors (Ryrs) on the opposing SR membrane. Muscle contraction is initiated by action potentials that depolarize the T tubules causing a conformational change in L-type calcium channels and a subsequent activation of RyRs, which release Ca2+ from the terminal SR. The Ca2+ signal is rapidly propagated via RyRs through the adjacent longitudinal SR to activate the contractile machinery. Recently, elegant experiments using mechanically skinned muscle fibres, where Ca2+ indicator dyes are ‘trapped’ in the T tubule system, have indicated that a fraction of Ca2+ released via Ryrs is extruded into the T tubules, and is quickly followed by Ca2+ influx via SOCs back into the muscle fibre . These data raised the possibility that reuptake of Ca2+ from the T tubule back into muscle cells by SOCs could comprise an integral part of skeletal muscle contraction–relaxation, by providing the means to sustain the SR load of Ca2+ required for persistent contractions.
In normal muscle fibres, STIM1 is localized to regions of the terminal SR that significantly overlap with RyR expression , and is either pre-localized to areas adjacent to ORAI1 or perhaps pre-coupled to the channel, and thus able to rapidly respond to the Ca2+-depletion induced conformational change that activates the channel. A recent study suggests that the C-terminus of STIM1 and STIM2 have particular abilities to bind to plasma membrane phosphoinositides . The synthesis of PtdInsP2 in vertebrate muscle is limited mainly to the T-tubule membrane [86, 87], and PtdInsP synthesis increases during rat myoblast differentiation . This localized synthesis of PtdInsP may serve to target the STIM1 C-terminus to the terminal SR membrane where it would be in position to rapidly activate ORAI1.
Myofibrils isolated from homozygous STIM1-deficient mice display essentially no SOCE and, when subjected to repeated depolarizations, show a rapid decrease in Ca2+ transient amplitude which is a hallmark of muscle fatigue . Thus, STIM1 appears to be required in mature muscle for sustained KCl-induced Ca2+ transients. This is supported by other studies which, in disrupting the triad structure, have at the same time disrupted SOCE, and have shown an increased propensity to muscle fatigue . In addition, increased muscle fatigue in ageing skeletal muscle is associated with decreased SOCE . Muscle fibres from heterozygous STIM1-deficient mice also showed an increased propensity to fatigue earlier than wild-type mice, suggesting that reduced STIM1 expression can disrupt SOCE in mature muscle fibres. The role of STIM2 in muscle development and function has not been explored.
Store-operated channels with similar characteristics to that of ICRAC have been described in various smooth muscle preparations, including vascular smooth muscle cells (VSMC) isolated from aorta [90, 91], portal vein  and pulmonary arteries , as well as smooth muscle cell (SMC) lines . The detailed biophysical characteristics of SOCs in SMCs have recently been reviewed . Much less is known about the regulation of these channels during smooth muscle differentiation. Unlike skeletal or cardiac muscle, whose cells are terminally differentiated, SMCs retain remarkable plasticity and can revert from a differentiated, contractile phenotype back to a de-differentiated, proliferative phenotype in response to local environmental cues.
STIM1 and ORAI1 expression has been demonstrated in VSMC from several species and reduction of STIM1 or ORAI1 expression shown to impair SOCE and whole-cell SOC currents [91, 96–99]. However, despite robust expression of STIM2, ORAI2 and 3 in VSMC cultures, knockdown of these genes appears to have little effect on VSMC SOCE [55, 91]. In proliferating rat VSMCs, SOCE mediated by STIM1 and ORAI1 has an unusual phenotype, being inhibited by low concentrations of 2-ABP, whereas in all other cells tested, low concentrations of 2-ABP potentiate SOCE. These observations suggest that the molecular composition of SOCs in VSMCs may differ to that of other cells, or, that unique, VSMC-specific post-translational modifications of ORAI1 or STIM1 contribute to the characteristics of SOCE in this cell type .
Although the role of STIM1 in VSMC differentiation has not been examined directly, several studies have noted that SOCE is increased in proliferating SMCs when compared with quiescent, more differentiated SMCs [91, 93, 96, 100], suggesting that down-regulation of SOCE is associated with the acquisition of a differentiated phenotype. Moreover, in a model of VSMC de-differentiation, STIM1 expression levels are rapidly up-regulated, concomitant with an increase in SOCE, consistent with a role for STIM1 in either the acquisition or maintenance of a de-differentiated, and therefore proliferative and migratory, VSMC phenotype. Indeed, depletion of STIM1 reduces the proliferative and migratory capabilities of de-differentiated VSMCs [91, 101–104]. A failure of these cells to progress to S phase of the cell cycle is associated with up-regulation of p21 and reduction in Rb phosphorylation  and a reduction in CREB phosphorylation  and NFAT transcriptional activity , suggesting that multiple signalling pathways may be regulated by STIM1 in VSMCs.
De-differentiation of VSMCs plays a major pathophysiological role in the development of atherosclerosis, restenosis following stenting or bypass surgery, and hypertension. In a model of neointima formation following balloon angioplasty, characterized by excessive proliferation of SMCs in the intima layer of the blood vessel, increased SMC proliferation is associated with increased STIM1 expression. Adenovirus-mediated delivery of shRNA directed to STIM1 in vivo following balloon injury significantly reduced neointima formation, suggesting that STIM1 could represent a viable therapeutic target to reduce restenosis following injury [101, 103], possibly utilizing surgical implantation of stents impregnated with drugs . Exercise can prevent VSMC proliferation and the associated up-regulation of STIM1 and TRPC1 in SMCs following stenting in swine . SOCE is elevated in freshly isolated preglomerular VSMC from hypertensive rats compared to normal rats , and STIM1 and ORAI1 expression is increased in vessels from spontaneously hypertensive rats , which have augmented vessel tone and are more sensitive to constrictor stimuli, key markers of arterial hypertension. This augmented phenotype could be abolished by intracellular delivery of antibodies targeting either STIM1 or ORAI1 . These studies demonstrate the important role of STIM1 in mediating pathological events associated with artherosclerosis and hypertension, and the potential for therapeutic intervention through modulation of STIM1 function or expression.
Early immunolocalization studies demonstrated STIM1 expression in the foetal and adult central and peripheral nervous system, particularly in Purkinge cells of the cerebellum . Reporter gene activity is also preferentially high in this region in STIM1 gene-trap mice , and high levels of STIM1 are detected in protein extracts from the cerebellum of mouse brains [37, 109]. However, a more recent study has demonstrated that mouse brain is the only tissue analysed where STIM2 is expressed much more highly than STIM1 . Comparison of SOCE in cortical neurons isolated from STIM1- and STIM2-deficient mice showed no defects in STIM1-deficient neurons, whereas STIM2-deficient neurons had lower basal Ca2+ levels and abrogated SOCE . These observations suggest a crucial role for STIM2 rather than STIM1 in regulation of SOCE in the nervous system. Ischemia-induced increases in intracellular Ca2+ play a critical role in triggering neuronal cell death and in STIM2-deficient neurons the increase in Ca2+ after in vitro ischemia is slower and neuronal recovery faster than in wild-type or STIM1-deficient neurons . This neuroprotective effect of STIM2 deficiency was also observed in vivo, with lower infarct volumes found in STIM2-deficient mice after an ischemic episode. Inhibitors of STIM2 function may thus have therapeutic value as neuroprotective agents to treat ischemic injury and potentially other neurodegenerative disorders involving Ca2+ homeostasis .
No gross neurological deficits or structural abnormalities were visible in STIM2-deficient mice, suggesting that reduced SOCE does not influence brain development in a major way . However, a pronounced cognitive impairment became apparent when mice were tested for hippocampus-dependent spatial memory using the Morris water maze test . Several Ca2+-sensitive signalling pathways operate to modulate hippocampus-dependent regulation of spatial learning and memory. CAMKIIα, activated at high levels of Ca2+/calmodulin activity, is required for optimal performance in spatial memory tests  and sustained CREB activation is additionally required for hippocampus-dependent spatial memory formation [111–113]. The effects of STIM2 deficiency on synaptic transmission and plasticity have not yet been explored.
The role of Ca2+ signalling in regulation of adipogenesis was indicated from observations of the obesogenic effects of the calcineurin inhibitor CsA in patients requiring immune suppression following transplant surgery, suggesting that the Cn/NFAT pathway negatively regulates adipocyte differentiation . The inhibitory effects of PGEF2α on adipogenesis were later found to be almost exclusively calcineurin-dependent  and adipocyte differentiation was enhanced by inhibiting calcineurin activity in mouse 3T3-L1 pre-adipocyte cells . We demonstrated that STIM1 expression increases progressively during differentiation of 3T3-L1 cells into mature, triglyceride-laden adipocytes, suggesting that either STIM1 contributes to the development of mature adipocytes from pre-adipocytes, or that the functionality of mature adipocytes requires STIM1 . Depletion of STIM1 in 3T3-L1 pre-adipocytes markedly enhances differentiation, associated with increased levels of CCAAT/enhancer-binding protein alpha (C/EBPα), one of the ‘master transcriptional regulators’ of adipocyte differentiation. Conversely, overexpression of STIM1, associated with an increased magnitude of SOCE, decreases triglyceride accumulation in a dose-dependent manner. C/EBPα expression is also abolished by STIM1 overexpression, suggesting that STIM1 negatively regulates adipocyte differentiation via a mechanism that is critical for the maintenance of C/EBPα expression.
The role of NFATs in regulation of adipogenesis in vivo was revealed by the phenotype of compound NFAT2-/NFAT4-deficient mice, which exhibit defects in fat accumulation, remain lean and are resistant to diet-induced obesity . An increased resting [Ca2+]i is characteristic of adipocytes from overweight human beings and rodents, which develops in as little as 3 weeks following initiation of a high fat diet in rats [117–120]. Manoeuvres which reduce these high [Ca2+]i can reverse weight gain, and improve insulin responses [121, 122], suggesting therapeutic manipulation of SOCE as a potential strategy in treatment of obesity.
The demonstrated role of STIM proteins in the physiology of the variety of cell types summarized above indicates that quantitative changes in STIM protein expression can influence proliferation, differentiation, and cell death in a cell specific manner, processes that are intimately involved in tumorigenesis. STIM1 maps to a region of 11p15.5 implicated in several childhood and adult tumours, including Wilms’ tumour, rhabdomyosarcoma, adrenal carcinoma, hepatoblastoma, bladder, breast, lung, ovarian and testicular cancers [123, 124], and was originally proposed as a candidate tumour suppressor in this region [125, 126]. Expression of STIM1 is barely detectable in rhabdomyosarcoma tumours . Tumour suppressor function is supported in studies showing that depletion of STIM1 expression transforms the weakly metastatic B16F0 mouse melanoma cell line into a more aggressive form, accelerating cell mobility in vitro and increasing metastasis in a mouse model of tumorigenesis . STIM2 maps to chromosome 4p15.1, a region implicated in invasive carcinomas of the head and neck, breast and lung [128–130].
An oncogenic rather than a tumour suppressive function has been demonstrated for STIM1 and STIM2 in glioblastoma multiforme, where both proteins have increased expression and/or increased copy number [131, 132]. The increased expression of both STIM1 and STIM2 in this tumour may correlate with the increased basal [Ca2+]i and SOCE measured in primary cultures from glioblastoma tumours, and in glioblastoma cell lines compared with normal human astrocytes , and potentially contribute to the highly malignant behaviour of this tumour type .
An oncogenic role for STIM proteins is also implicated in other cancers. Overexpression of STIM1 increases the migration of MCF-10A cells, a non-tumorigenic mammary gland cell line which expresses moderate endogenous levels of STIM1 . Conversely, siRNA-mediated depletion of STIM1 in the tumorigenic breast cancer cell line MDA-MB-231, which expresses higher levels of STIM1, reduces migratory capacity in vitro and dramatically decreases their ability to metastasize in vivo. Our own studies have demonstrated that overexpression of either STIM1 or STIM2 in rat PC12 pheochromocytoma cells does not affect the size of solid tumours formed after subcutaneous injection into nude mice, but causes significant invasion of cells into the body wall skeletal musculature, which was not evident after injection of parental PC12 cells (Fig. 3; M. Dziadek, unpublished results).
Figure 3. Haematoxylin and eosin stained sections of subcutaneous tumours derived from STIM1 or STIM2 overexpressing PC12 cells injected into nude mice. Control (vector) PC12 tumour cells (t) are aligned against connective tissue and skeletal muscle fibres (m) of the body wall, with only very limited infiltration of cells between muscle fibres seen in two of nine tumours. In STIM1 (n= 9) and STIM2 (n= 9) tumours, invasive tumour-derived cells have infiltrated between skeletal muscle fibres (m), with extensive infiltration evident in >50% of tumours of both types.
Download figure to PowerPoint
MDA-MB-231 cells in which STIM1 levels were depleted had decreased focal adhesion turnover, resulting in larger focal adhesions, and consequently stronger adhesion, implicating STIM1-mediated SOCE in regulation of components of the focal adhesion machinery. Overexpression of focal adhesion kinase, the critical regulator of focal adhesion turnover, is associated with cancer grade in some, but not all, cancers, suggesting that focal adhesion kinase may play alternative roles in different tumours and/or in different stages of tumour progression . Because STIM1 and STIM2 likely regulate a diverse number of downstream signalling cascades, the pathways and partners involved in STIM1- and STIM2-mediated cancer progression are likely to be tumour cell type dependent. Further studies on the role of STIM proteins in different cancers are now required to determine in which cases therapeutic agents designed to either increase or decrease the activity of STIM1 and STIM2, and thus SOCE, would be useful for cancer treatment.