A wide variety of cellular function depends on the dynamics of intracellular Ca2+ signals. Especially for relatively slow and lasting processes such as gene expression, cell proliferation, and often migration, cells rely on the store-operated Ca2+ entry (SOCE) pathway, which is particularly prominent in immune cells. SOCE is initiated by the sensor proteins (STIM1, STIM2) located within the endoplasmic reticulum (ER) registering the Ca2+ concentration within the ER, and upon its depletion, cluster and trap Orai (Orai1-3) proteins located in the plasma membrane (PM) into ER-PM junctions. These regions become sites of highly selective Ca2+ entry predominantly through Orai1-assembled channels, which, among other effector functions, is necessary for triggering NFAT translocation into the nucleus. What is less clear is how the spatial and temporal spread of intracellular Ca2+ is shaped and regulated by differential expression of the individual SOCE genes and their splice variants, their heteromeric combinations and pre- and posttranslational modifications. This review focuses on principle mechanisms regulating expression, splicing, and targeting of Ca2+ release-activated Ca2+ (CRAC) channels.
since the discovery of the STIM and Orai proteins as the major molecular components underlying store-operated Ca2+-entry (SOCE) in 2005 and 2006, investigations into their mechanisms of activation, regulation, and physiological roles have continuously increased and detailed reviews have accompanied major developments. Much of our understanding has been gained from mutations in critical domains of the proteins, several of which lead to disease phenotypes found in patients, and from targeted often tissue-specific deletions in mouse models. Many recent reviews have covered structural and functional aspects of STIM- and Orai-gated SOCE (11, 13, 14, 28, 29, 48, 57, 58, 60, 62, 64, 72, 79), with much knowledge gained form mutational analysis and loss of function mutations seen in immunodeficient patients. Less well understood and only recently discovered are a number of gain-of-function mutations of STIM1 and Orai1 which can lead to pathologies such as Stormorken syndrome, nonsyndromic tubular aggregate myopathy (TAM), and York syndrome (reviewed in Ref. 41). In general, it is much less clear if and how smaller changes in expression levels of the two STIM genes and three Orai genes and their splice variants shape cell-specific cellular responses. Given the fact that the stoichiometry of the STIM1:Orai1 protein ratio is a major determinant in determining current size and inactivation properties (31, 40, 71) and that local domains of calcium entry are able to selectively activate different NFAT transcription factor isoforms (38, 87), regulating expression levels of the SOCE genes themselves is likely to impact cellular responses. Figure 1 depicts several checkpoints of principle regulatory mechanisms that affect the expression of STIM and Orai genes and the trafficking and localization of the corresponding proteins and will be discussed in this review. Of possible importance are also epigenetic regulatory mechanisms; in one of the original manuscripts identifying STIM1, downregulation of the STIM1 promoter by in vitro CpG methylation was described (66), and it will be interesting to see whether this also plays a role in vivo.
Regulation of Promotor Activity
Reduced STIM1 and Orai1 activity results in a number of severe immune deficiencies, muscle pathologies, deficient enamel production among others (41), and abrogated activation of the transcription factor NFAT, but few studies investigated how regulation of SOCE gene promoters themselves may affect STIM and Orai function by regulating their gene expression levels. Database mining (e.g., http://www.sabiosciences.com) predicts binding sites for multiple transcription factors such as Pax-4a, c-Rel, CUTL1, FOXO1, RelA, NF-κB, and RSRFC4 for Orai1 and AP-1, NF-κB, SRF, c-Fos, c-Jun, hepatic nuclear factor 1 (HNF1), and POU2F1 for STIM1. Orai2, Orai3, and STIM2 also have a number of predicted sites. While some SOCE genes have similar predicted transcription factor-binding sites (e.g., NF-κB for both STIM1 and Orai1), most of the promoter regions show potential binding sites for different transcription factors. Experimental evidence for regulation of SOCE promoter activities is scarce. Lang's group (22, 43) investigated the effects of serum and glucocorticoid-inducible kinase SGK1, a serine/threonine-protein kinase that is involved in the regulation of a wide variety of ion channels, membrane transporters, cellular enzymes, and transcription factors, among others, on the activity of SOCE. They showed that decreased levels of STIM1 and Orai1 transcripts detected in SGK1−/− cells were due to decreased NF-κB activity. Silencing of NF-κB subunits p65, p50, or p52 or application of the NF-κB inhibitor, Wogonin, decreased Orai1 and STIM1 transcript levels while NF-κB subunit overexpression increased transcripts. Chromatin immunoprecipitation and luciferase assays defined NF-κB-binding sites in the respective promoter regions, revealing genomic regulation of Orai1/STIM1 by SGK1-dependent NF-κB signaling (22). Tiruppathi's group investigated the basis of the increased expression of STIM1 in endothelial cells (EC) seen during sepsis (17). Sepsis is a life-threatening complex systemic inflammation caused by massive bacterial infection and the concomitant release of lipopolysaccharides (LPS), also known as endotoxins or lipoglycans derived from the bacterial walls. Addition of LPS induced STIM1 mRNA, which could be prevented by inhibition of either NF-κB or p38 MAPK activation by pharmacological agents. Accordingly, silencing of the NF-κB proteins (p65/RelA or p50/NF-κB1) or the p38 MAPK isoform p38α prevented LPS-induced STIM1 expression and increased SOCE in ECs. NF-κB- and AP1-binding sites were identified in the 5′-regulatory region of human and mouse STIM1 genes, which are increasingly occupied upon LPS stimulation. Silencing of c-Fos, but not c-Jun, markedly reduced LPS-induced STIM1 expression in ECs. In addition, silencing of p38α prevented c-Fos expression in response to LPS in ECs, confirming that p38α signaling can mediate the expression of c-Fos (8). These results support the hypothesis that cooperative signaling of both NF-κB and AP1 (via p38α) enhances STIM1 expression in ECs. This increased expression may contribute to the lung vascular hyperpermeability response during sepsis (17).
In a more specialized cell type, glomerular mesangial cells (MC), high glucose induces an increase in SOCE and in STIM1 protein abundance (12). This effect is indirectly mediated by disabling hepatic nuclear factor-4α (HNF4α), a transcription factor that responds to high glucose by repressing the transcription of STIM1. Experiments in MC cells showed that knockdown of HNF4α significantly upregulated mRNA expression levels of STIM1 and Orai1 and protein expression of STIM1, while overexpression of HNF4α reduced protein levels of STIM1 in HEK293 cells (83). High glucose leads to an increased STIM1 expression by impairing binding of HNF4α to the STIM1 promoter, thus releasing repression of STIM1 transcription by HNF4α. Since the STIM1-gated store-operated Ca2+ entry pathway has an antifibrotic effect, inhibition of HNF4α in MCs might be a potential therapeutic option for diabetic kidney disease. What is unclear is why the increase seen in Orai1 mRNA abundance fails to increase protein abundance and whether this is a direct or indirect effect of altered HNF4α activity.
Although regulation of transcript levels of SOCE genes by transcription factors does not represent an immediate regulatory mechanism, identifying master regulators of SOCE gene promoter activities will aid in identifying developmental stages and physiological as well as pathophysiological conditions in which SOCE proteins are more likely to be critically important. Given the number of potential regulating transcription factors, it is clear that much experimental work is ahead.
Alternative splicing is a common mechanism to diversify functional aspects of many proteins including calcium conducting ion channels and is especially well studied for the multiple exon encoded voltage-gated calcium channel genes (45, 73). Orai channels by comparison are encoded by small genes with the coding region of all human Orai genes encoded by only two exons. Human Orai2 has an alternative exon located within the 5′-untranslated region (UTR) of the gene, splicing of which can lead to an alternatively used methionine. The predicted resulting protein, however, would lack the first trans-membrane region thereby representing a nonfunctional channel. Whether or not this short protein variant exists in vivo is unclear. Wissenbach et al. (85) investigated the genomic structure of all three murine Orai genes and found two genomic loci for Orai2 and evidence for an mRNA with an 5′-extension of 14 amino acids compared with the human homolog. No alternate variants are listed for human Orai3, located on chromosome 16. Alternate translation of human Orai1 will be discussed below. Compared with the Orai genes, STIM genes cover larger genomic regions and are encoded by multiple exons, with a highly conserved exon structure. Human STIM1, located on chromosome 11, is encoded by 12 exons in its longest form (STIM1L) (16, 67). The main difference between STIM1 and STIM2 lies in the length of exon 11, which encodes for 67 nucleotides in STIM1 and the corresponding exon in STIM2 encoding 274 nucleotides (see below). Darbellay et al. (16) discovered that alternative splicing and extension of exon 11 leads to a longer STIM1 protein, STIM1L, expressed in adult human muscle fibers and myotubes differentiated in vitro. The insertion adds 106 residues and an actin-binding domain into the cytosolic region of STIM1. STIM1L was originally proposed to form permanent clusters with Orai1, allowing more rapid activation especially needed during repetitive stimulation. While Horinouchi and coworkers (32) confirmed the existence of STIM1L in skeletal muscle and found it to inhibit TRPC3 and TRPC6 activity, Luo et al. (47) found STIM1L protein in neonatal rat cardiomyocytes, with decreasing expression during postnatal cardiac development. Upon agonist- or afterload-induced cardiac stress, STIM1L expression remerged. Recently, Sauc et al. (67) investigated the molecular mechanism of STIM1 vs. STIM1L gating in more detail. While STIM1 was able to expand cortical ER cisternae, STIM1L was unable to do so, but could trap and gate Orai1 channels without remodeling cortical ER. Interestingly, STIM1 but not STIM1L was found to support fast activation or SOCE (67). Besides the longer isoform STIM1L, an additional splice variant of STIM1 exists (NM_001277962.1), which contains an alternate exon and uses an alternate splice site in the 3′-coding region, resulting in a frame-shift and premature translational stop. The resulting predicted protein (isoform 3) has a distinct C-terminus and is shorter by 145 residues, compared with STIM1, but its role in vivo has not been investigated. Defective splicing of STIM1 has also been discovered in a child with fatal classic Kaposi sarcoma (KS). The resulting defective protein production and the complete absence of SOCE in a patient-derived B-cell line most likely precipitated the development of lethal complications after viral infection (10).
The genomic region of STIM2 shows a related organization, the common full-length human STIM2 being encoded by 12 exons. This variant (STIM2.2 or STIM2α) lacks exon 9, and the resulting protein, owing to its reduced EF-hand Ca2+ affinity compared with STIM1, is implicated in the regulation of basal Ca2+ homeostasis (9). The role of STIM2 and STIM1 in activation or inhibition of SOCE has been reviewed in Refs. 3, 34, 62, 75. STIM2 shows prominent expression within brain tissues where the protein contributes to capacitive Ca2+ entry in neurons and plays a key role in hypoxic neuronal cell death (5) with functional implications also for neurodegenerative disease (48, 55, 61, 65, 89). Recently, we (53) and the Lewis group (63) identified expression of a longer STIM2 splice variant which includes a novel exon 9 (STIM2.1 or STIM2β; Fig. 2), thus resulting in a gene encoded by 13 exons and shifting the numerals of the downstream exons. Although exon 9 only encodes for 8 amino acid residues (VAASYLIQ for H. sapiens), these are highly conserved and are inserted into the channel-activating domain [CAD or STIM-Orai-activating regions (SOAR)]. The resulting protein STIM2.1 by itself is indeed unable to tightly bind and to activate Orai channels. While STIM2.1 shows a reduced expression compared with STIM2.2, we found relevant expression (nearly equal to STIM2.2) in naïve CD4+ and CD8+ T cells with significant downregulation of expression upon differentiation into effector cells (53), whereas Rana et al. (63) found upregulation of the STIM2β (2.1) over STIM2α (2.2) upon early myogenic differentiation of murine C2C12 myoblasts. In naïve T cells, silencing of STIM2.1, but not of STIM2.2, caused an increase in SOCE, indicating a dominant-negative function of STIM2.1 in vivo. Whether the dominant-negative function of the spliced-in residues of STIM2.1 in combination with STIM2.2 or with STIM1 measured upon heterologous coexpression are due to indirect effects caused by STIM2.1 blocking STIM2.2 from accessing the Orai1 C-terminal binding domain, possibly exacerbated by STIM2.1's increased Ca2+-dependent affinity towards calmodulin (53), or by a more active inhibitory contribution of the eight residues as proposed by Rana et al. (63) remains to be clarified. Given the existence of this inhibitory splice variant, the often small effects seen upon genetic deletion of STIM2 may indeed underestimate a STIM2.2-mediated effect, as the concomitant deletion of STIM2.1 deletes the negative regulator. How splicing is regulated and which of the many regulatory splicing factors (80–82) is determining the ratio of STIM2.1/STIM2.2 is unknown.
Similar to STIM1, database mining also predicts a cDNA resulting in a 147 amino acid residue shorter STIM2 variant (STIM2.3) compared with STIM2.2 due to use of an alternate exon 13 (Fig. 2) that contains a short unique 3′-UTR. However, its expression in vivo has not yet been confirmed or investigated.
Given the number of exons for both STIM1 and SITM2, other splice variants are also predicted (http://www.ensembl.org) but mostly lack the NCBI reference annotation.
RNA editing is a relatively rare event, but it can lead to functionally different receptors or ion channels. During RNA editing, specific nucleotide sequences within the RNA molecule can be exchanged, deleted, or modified, processes that can occur in the cell nucleus and cytosol but also within mitochondria. A beautiful study, by the late David Yue, found that RNA editing of the IQ domain of voltage-gated Cav1.3 (CACNA1D) modulates their Ca2+-dependent inactivation (36). Whether or not RNA editing modulates SOCE genes is unknown.
RNA Stability: miRNA versus RNA Stabilizing Factors
In contrast to RNA editing, modulation of RNA stability by RNA silencing through miRNA molecules appears to be a rather common mechanism. Micro RNAs (miRNA) are small noncoding RNA molecules (containing ∼22 nucleotides) that are encoded by eukaryotic nuclear DNA in plants and animals and affect the stability of target mRNAs by complementary base pairing (4). In their mechanism of action, miRNA resemble the action of small interfering RNAs (siRNA) although their silencing efficiencies and target specificities are usually much lower (42). A typical miRNA often has ∼400 target sequences, usually located within the 3′-UTR of the target mRNA. By targeting cognate mRNAs, miRNA binding can result in less efficient translation into proteins at the ribosomes, in cleavage of the target mRNA or in destabilization of the mRNA (4). miRNAs are involved in regulating most, if not all, biological processes in different cell and tissue types, and alterations of miRNA expression profiles have been implicated to be involved in carcinogenesis, inflammation, and many different diseases (summarized in Ref. 23). Zare et al. (88) recently defined an evolutionarily biased distribution of miRNA sites towards regulatory genes with high promoter-driven intrinsic transcriptional noise; although STIM1 or Orai1 have not emerged as hits, STIM2 appears as a gene with high transcriptional noise.
While certain miRNAs in combination with RNA-binding proteins turn out to be oncogenic drivers [e.g., miR-let-7 in combination with LIN28 for cancer stem cells (92)], many miRNAs in combination with their target proteins are likely passengers, fine-tuning cellular responses and modulating oncogenic potential. The driver-passenger concept also applies for the STIM and Orai proteins, whose up- or downregulation has been implicated to play a role for carcinogenesis and has recently been reviewed by Hoth (33).
Only a handful of studies have investigated the effects of miRNA on the expression of SOCE genes. Two studies have studied the effects of miR-185 on STIM1. In both studies, miR-185 significantly downregulated expression of STIM1 with resulting effects on migration of colorectal cancer cells and microvascular endothelial cells being reverted either by mutation of the miR-target site or by overexpression of STIM1 (33, 34). MiR-195 has also been found to downregulate STIM1 and to decrease migration of intestinal epithelial cells. miR-195 binding to STIM1 was counteracted by the RNA-binding protein HuR associating with the STIM1 3′-UTR and thereby prevented STIM1 mRNA decay (93). The above mentioned studies were conducted in cell lines. However, STIM1 overexpression has been observed, among other cancers, in a subset of colorectal cancer (CRC) patients and associated with cancer cell invasion and migration, and a recent report correlated clinical features and genetic profiles of STIM1 in colorectal cancers. Here, the authors identified 11 prognostic mRNA/miRNA predictors associated with the overall survival of colon adenocarcinoma patients, suggesting a correlation of STIM1-associated features to clinicopathological outcomes (86). STIM1 is also indirectly affected by expression of miR-424 in vascular smooth muscle cells. Although STIM1 expression was decreased and SOCE reduced upon overexpression of miR-424 or its rat orthologoue miR-322, reporter assays indicated this was an indirect effect (52).
The microRNA miR-519 has been shown to inhibit cell proliferation, triggering senescence and decreased tumor growth. Using biotinylated miRNA as a probe, both Orai1 and ATP2C1, a Ca2+-ATPase located in the secretory pathway with dysregulation implicated in cancer (15), among other targets have been identified, and expression of miR-519 aberrantly elevated levels of cytosolic [Ca2+] storage in HeLa cells. Whether and how the downregulation of Orai1 contributed to this effect is unclear.
Clearly, given the number of predicted miRNA-binding sites within the 3′-UTR of all five genes, much more work is ahead. Because of the promiscuity of any given miRNA, the challenge is on proving relevant, reproducible, and specific effects on SOCE. In addition to the complexity of multiple potential targets, the significant variability in miRNA expression profiles of primary hematopoietic cells seen between different donors raises further concern (70).
Table 1 summarizes the results of transcriptional regulators and identifies miRNAs of SOCE genes.
The preceding paragraph described how RNA elements (miRNA and RNA protein-binding sites) within the mRNA can affect the translation efficiency, which is defined as the rate of mRNA translation into protein (69). Schwanhausser et al. (69) measured absolute levels of RNA and protein abundance and turnover for more than 5,000 mammalian genes and found that cellular abundance of proteins is predominantly controlled at the level of translation.
STIM1 and STIM2 are type I trans-membrane proteins and have their N-terminal domains targeted to the ER lumen. While STIM1 contains a conventional short signal peptide with 22 residues, mammalian STIM2 proteins contain an unusually long N-terminal region with the initial 87 amino acids encoded by exon 1, which precede 14 residues encoded by exon 2 that are recognized as a signal peptide, if translation is initiated at Leu88 (25). Analyses of several mutations around a putative noncanonical translational start codon and N-terminal (Edman) protein sequence analyses of the mature protein led to the initial conclusion that STIM2 translation is not initiated with an AUG codon, but instead at UUG corresponding to Leu88 (84). Graham et al. (25) investigated the long N-terminal sequence, which is highly conserved in placental mammals, and found that three different STIM2 proteins can be detected: a noncleaved protein containing the full 101 N-terminal sequence remaining within the cytosol (834 aa), a cleaved version (733 aa) requiring cleavage of the full signal peptide (1–101) and corresponding to the protein that is generated with addition of the artificial STIM1 sequence (9), and an additional cleavage product of the 101 residue N-terminal domain that would result in a signal peptide fragment (SPF). The SPF fragment is released into the cytosol and has been implicated to regulate NFAT- and NF-κB-mediated gene transcription (25). However, whether these cleavage products are of functional relevance in vivo still needs to be determined. In our hands, we did not find functional differences between STIM2.2 with the long preceding signal peptide or containing the short STIM1 derived signal peptide (53).
Another aspect of translational regulation for Orai1 has been identified by Putney's group: Here alternative use of two different methionines as translation initiators give rise to two different Orai1 isoforms with distinct mobilities within the plasma membrane (24). A recent study showed that the longer mammal-specific variant (Orai1α) exhibited stronger Ca2+-dependent inhibition and also is the variant needed to support the arachidonic acid-regulated current Iarc, which was not supported by the shorter Orai1β (19). The clear implication of this study is that Iarc must have distinct properties between mammals and other species, or might not exist in cells of nonmammalian origin.
Sorting: Complex Glycosylation, Targeting
Both STIM1 and STIM2 proteins contain luminal N-linked glycosylation sites that receive basal mannose rich glycan chains within the ER lumen. STIM1 contains two and STIM2 only one site, glycosylation of which remains EndoH sensitive, indicating that the majority of the proteins remain within the ER (84). While STIM2 contains a classic C-terminal ER retention motif (KKXX), STIM1 lacks two residues of this motif (21). Interestingly, a small amount of STIM1, but not of STIM2, can translocate to the plasma membrane, where it is implicated in activation of store-independent arachidonate-regulated Ca2+ (ARC) channels (78). Alterations of STIM1's two luminal N-glycosylation sites can lead to decreased (54) or increased STIM1 function that can be explained by altered rates of STIM1 oligomerization (40). Most channel proteins located in the plasma membrane contain N-linked glycosylation sites that facilitate trafficking to the membrane or retain proteins at the surface by limiting retrieval. Orai proteins are small proteins with four transmembrane domains and relatively short extracellular loops. Of the three paralogues, only Orai1 contains a single extracellular N-glycosylation motif (N223 for human Orai1). This site endows Orai1 with peptide-N-glycosidase F (PNGase F)-sensitive but EndoH-resistant complex oligosaccharides (26, 40), resulting in a significant shift in its apparent molecular mass. Expression of the short Orai1 variant described above would yield a glycosylated channel with a reduced molecular mass. Mutation of the Orai1 N223 site appeared to yield larger SOCE when expressed in fibroblasts of SCID patients (26) but had not been further investigated. We found Orai1's glycotype to be complex and EndoH resistant (40), decorated with terminal sialic acids and highly cell type specific, giving rise to very different Orai1-specific molecular masses in Western blots of different primary cells and cell lines. While Orai1 N223A mutants show normal function in HEK293 cells (40), glycosylation-deficient Orai1 increased SOCE in T cells. The functional significance of cell type-specific glycosylation patterns of Orai1 depended on the nature of its terminal sialic acids and on the expression of sialic acid-binding partners, most likely determining the degree of surface retention within the plasma membrane (19a).
Neither Orai2 nor Orai3 proteins contain N-glycosylation consensus motifs, and expression in stable STIM1-expressing cells yields much smaller Ca2+ entry and currents compared with Orai1 (2, 18, 26, 46, 51). Using surface biotinylation experiments, ∼ 40% of Orai1 overall protein is localized in the plasma membrane (2, 27, 40), but only ∼14% for Orai3 and ∼10% for Orai2. Silencing or overexpression of Orai1 had a significant impact on the amount of surface-expressed Orai3, but not of Orai2 localized to the plasma membrane. A similar facilitation of Orai3 localization by Orai1 is also seen upon formation of an artificial immunological synapse (IS) in T cells or upon store-depletion and colocalization to STIM1 clusters (2). Some of the reduced surface expression of Orai3 may be due the absence of glycosylated residues; however, improving the Orai3 C-terminal STIM1 interaction site abrogated Orai3's functional and geographic dependency on Orai1 (2). While Orai2 shows Orai1 independency in localization studies (to STIM1 clusters or towards the IS), it reduced overall SOCE amplitude when overexpressed or increased SOCE amplitude upon knockdown in Jurkat T cells (2). Similar results concerning the role of Orai2 for SOCE were obtained in OUMS-27 cells derived from human chondrosarcoma (37). The latter studies applied a number of further fluorescence resonance energy transfer (FRET) and bimolecular fluorescence complementation (BifC) analysis methods that indicate that Orai1 and Orai2 form heterotetramers. Heteromultimerization with either Orai2 or Orai3 thus reduces Orai1 function and while Orai3 adds redox insensitivity to the complex (2, 7, 30) and enables heteromeric complexes to function in a non-store-operated manner gated by 2-APB, arachidonic acid, or leukotriene C4 (20, 56, 59, 68, 90), it remains to be seen which additional properties or features Orai2 adds to the SOCE complex. Clearly, questions regarding trafficking and subunit composition of the three Orai paralogues and their variants remain enigmatic.
Protein Turnover and Degradation
Machaca's group (27) found that Orai1 recycles rapidly at the plasma membrane (Kex ≃ 0.1 min−1), with a subset of intracellular Orai1 localized to a sub-plasmalemmal compartment. Store depletion enriched Orai1 in the plasma membrane in a STIM1-dependent fashion. Upon high STIM1 expression, Orai1 is trapped into intracellular STIM1 clusters (43), ultimately leading to degradation of Orai1 protein and a concomitant change in the STIM1:Orai1 ratio that can lead to increased Orai1-mediated currents (40). In murine fibroblasts (NIH3T3 cells), the absolute levels of RNA and protein abundance and turnover were determined for more than 5,000 mammalian genes. Although the Orai proteins were not among the ones investigated, STIM1 was found with an average of 53,361 protein molecules per cell, an average of 15 mRNA molecules, a protein half-life average of 43 h, and transcription rate average of 0.95 molecules/(cell*h) (69). Concerning the number of molecules per cell, STIM1 was close to the median.
If and how turnover rates and stability of STIM1-Orai1 clusters are affected by binding of the cytosolic Ca2+-dependent proteins such as the short splice variant of CRACR2A in T cells (77), calmodulin, caveolin, or SPCA2 (76) is unclear. Regulation of Orai1 by differential glycosylation (see above) also will most likely affect its protein turnover.
Intracellular protein degradation can be achieved in several ways, e.g., adding an ubiquitin flag to proteins and thereby targeting these proteins for proteasome degradation, or proteolysis within lysosomes. While the ubiquitin pathway is selective, the autophagy-lysosomal pathway usually is not.
Orai1 has been identified as a target of Nedd4-2, a ubiquitin ligase which covalently tags plasma membrane proteins for degradation (43). Nedd4-2 is activated by the energy-sensing, AMP-activated kinase AMPK, accelerating Orai1 degradation, but phosphorylation of Nedd4-2 by the serum- and glucocorticoid-inducible kinase SGK1 is able to prevent interaction of Nedd4-2 with Orai1 by targeting Nedd4-2 to 14-3-3 proteins (43). In confirmation of the above mentioned findings, T cells from AMPKα−/− animals show an increased abundance of Orai1 within the plasma membrane (6). In addition, Orai1 has also been found to interact with ubiquilin 1, promoting its ubiquitination as well as lysosomal degradation (44). We found a Ca2+-dependent increased protein degradation of Orai1 upon co-overexpression with a STIM1 gain-of-function mutation (40). Whether increased degradation of Orai1 leading to a concomitant alteration and possible optimization in the STIM1:Orai1 protein ratio is also an additional cause of the increased SOCE seen in patients with Stormorken or TAM syndromes due to mutations within STIM1 (see above) is unclear. STIM1 has also been shown to be ubiquitinated in hippocampal neurons, reducing its surface expression. While proteasome inhibitors enhanced surface expression and SOCE, overexpression of a different E3 ubiquitin ligase (POSH) reduced surface expression of STIM1 (39).
Modulation of STIM and Orai gene expression and transcript levels is clearly an emerging field and will aid in understanding why SOCE signals in cells can be very different. Heterologous coexpression of the paralogues Orai2 and Orai3 reduce Orai1-mediated SOCE and can alter its redox sensitivity (Orai3), but the mechanisms and the rules for hetero- or homomeric assembly with themselves or with Orai1 and their contribution to the physiology of cells are still very much an open field (34). Careful analysis of Orai2 and Orai3 deletion models will certainly aid to identify their contributions in the future. Many of the mechanisms discussed above (transcription factors, miRNA, glycosylation) will contribute to different amounts depending on developmental stage, cell type, and expression of the cognate lectin. Most likely, these modifications will result in smaller changes in SOCE, but they may be sufficient to predispose individuals for developing pathological conditions, e.g., allergies or autoimmune disease, or to accelerate or reduce migration of cancer cells. Regulation of STIM splicing certainly represents a very attractive means to significantly affect SOCE function during certain developmental stages or to lock cells in a mode of reduced or altered capacity to become activated.
While this review included glycosylation as posttranslational modification, other modifiers such as phosphorylation, oxidative modifications, glutathionylation, and pH effects have not been discussed here, but see Refs. 49, 50, 62, 74. Both STIM1 and STIM2 have multiple phosphorylation sites and much is still to be learned about each of their regulatory mechanisms.
B. A. Niemeyer acknowledges funding from the Deutsche Forschungsgemeinschaft (DFG): SFB894 A2, SFB1027 C4, and FOR2289 P6.
No conflicts of interest, financial or otherwise, are declared by the author.
B.A.N. prepared figures; B.A.N. drafted manuscript; B.A.N. edited and revised manuscript; B.A.N. approved final version of manuscript.
↵* This review article is part of a Theme series: STIM and Orai Proteins in Calcium Signaling.
- Copyright © 2016 the American Physiological Society