Control of STIM and Orai function by post-translational modifications

Summary

Store-operated calcium entry (SOCE) is mediated by the endoplasmic reticulum (ER) Ca²⁺ sensors stromal interaction molecules (STIM1 and STIM2) and the plasma membrane Orai (Orai1, Orai2, Orai3) Ca²⁺ channels. Although primarily regulated by ER Ca²⁺ content, there have been numerous studies over the last 15 years demonstrating that all 5 proteins are also regulated through posttranslational modification (PTM). Focusing primarily on phosphorylation, glycosylation and redox modification, this review focuses on how PTMs modulate the key events in SOCE; Ca²⁺ sensing, STIM translocation, Orai interaction and/or Orai1 activation.

I. Introduction

Calcium (Ca²⁺) signaling is important for numerous cellular functions and is regulated by many pathways and channels, including store-operated Ca²⁺ entry (SOCE). Stromal interaction molecules (STIMs) and Orai’s are membrane proteins that serve as required components of SOCE. SOCE occurs in response to endoplasmic reticulum (ER) Ca²⁺ depletion; STIM1 and STIM2 sense changes in ER Ca²⁺ content, responding by translocation within the ER towards sites of close apposition with the PM, where they can physically couple with members of the Orai family of Ca²⁺ channels [1, 2]. SOCE is a common signaling event in both excitable and non-excitable cells, initiated as a component of a wide range of physiological and pathophysiological responses [3]. Given the breadth of different responses that this signaling system can contribute to, it is perhaps not surprising that STIM and Orai function can be modulated by a wide range of different types of post-translational modifications (PTMs). Here we will describe both how the process of store-operated Ca²⁺ entry occurs and how PTMs can modify these responses.

Best known as an ER Ca²⁺ sensor, STIM proteins can detect a range of cellular stressors including oxidative stress, temperature change and hypoxia [4]. Human STIM1 contains two homologues (STIM1 and STIM2) that are both single-pass transmembrane proteins. STIM1 serves as the primary mediator of SOCE while STIM2 is a ubiquitously expressed [5] regulator of basal Ca²⁺ content [6] and a facilitator of receptor-operated Ca²⁺ entry [7–10]. However, STIM2 also participates in receptor activated Ca²⁺ signaling event with STIM1 and enhances the sensitivity of STIM1 to decreased agonist concentrations [10–12]. STIM1 and STIM2 share multiple homologous regions, notably the EF-hand, sterile a motif (SAM) domain, and three coiled-coil (CC) domains (Fig 1A) [4, 5, 13]. However, modest structural differences within these domains lead to functional differences that differentiate these 2 isoforms. For example, the STIM2 EF-hand has a lower affinity for Ca²⁺, causing it to be constitutively active under resting conditions [14]. Importantly, the amount of constitutive activity is limited due to a combination of its lower expression level [5] and structural differences in the SAM domain [6], the N-terminus [15] and within the STIM-Orai interaction region (SOAR) [16, 17]. Additionally, STIM1 and STIM2 diverge structurally in their C-terminus with STIM1 containing a proline-serine (P/S) rich domain while STIM2 contains a non-homologous proline-histidine (P/H) rich domain [4].

Figure 1: Domain architecture of STIM/Orai proteins.

Schematic diagrams of the domain structure of each member of the STIM and Orai families, with key amino acid residues discussed within this manuscript as demonstrated or potential targets for post-translational modification marked. (A) STIM1 (685 amino acids) and STIM2 (833 aa) are single membrane spanning proteins with luminal/extracellular N-termini (when localized to the ER or PM respectively) and cytosolic C-termini. The N-terminus incorporates a signal peptide (SP), canonical and hidden EF-hands (c/he) and the sterile α-motif (SAM) upstream of a single transmembrane domain (TM). The C-termini of both STIM1 and STIM2 contain a coiled-coil domain (CC1), the STIM-Orai activating region (SOAR), an inhibitory domain (ID) responsible for Ca²⁺-dependent inhibition of Orai1 function and a poly-lysine (K) region at the distal C-terminus responsible for association with the PM. Unique to STIM1 is a proline-serine (P/S) rich domain, while a non-homologous proline-histidine (P/H) rich region is found in STIM2. (B) Orai1 (301 amino acids), Orai2 (254 amino acids) and Orai3 (295 amino acids) are tetra spanning proteins found primarily on the plasma membrane (PM). The location of each TM is marked, as is the unique proline-rich domain (P) of Orai1.

Although primarily found within the ER, 5 to 10% of STIM1 (but not STIM2) is found in the plasma membrane (PM) [18–20]. This difference is the result of a di-lysine ER retention sequence within the C-terminal domain of STIM2 only [18]. Interestingly, STIM1 was initially identified as a surface protein on pre-B lymphocyte that interacts with stromal cells in bone marrow and contributes to their differentiation into B cells [13, 21]. There is little reason to believe that PM-STIM1 contributes to SOCE, as the possibility that translocation to the PM contributes to SOCE is not believed to occur [20, 22]. Whether or not PM-STIM1 functions as a modulator of SOCE or is involved in unrelated processes has yet to be established.

The Orai family of Ca²⁺ selective ion (also known as Ca²⁺ release activated Ca²⁺; CRAC) channels are found primarily within the PM and serve as the pore-forming unit of SOCE [1, 23]. Humans encode three Orai homologues (Orai1, Orai2, Orai3); although ubiquitously expressed, Orai1 serves as the major family member found in most cell types while Orai2 and Orai3 are expressed at higher levels in the brain [1, 24, 25]. All three homologues contain four transmembrane regions and a CC domain at the C-terminus [3, 25, 26], however, there are a number of differences in sequence in the N- and C-termini as well as the 2,3 and 3,4 loops (Fig 1B) [13]. All 3 Orai channels are believed to form hexamers, with TM1 serving as the pore lining domain and the C-terminus interacting with STIM [27]. Although, all three Orai homologues can function in SOCE upon Ca²⁺ store depletion, Orai1 demonstrates Ca²⁺ currents ~2–3-fold greater than Orai2 and Orai3 [25, 28]. Consequently, the presence of Orai2 and/or Orai3 ultimately leads to decreased SOCE, while also facilitating graded agonist responses [29–31].

When ER Ca²⁺ store depletion occurs, Ca²⁺ dissociates from STIM1 and undergoes a conformational change [13]. STIM then will migrate to ER-PM junctions, which can be visualized as “puncta” in light microscopy [32]. STIM multimers then associates with Orai1 at the ER-PM junction through the STIM-Orai activating region (SOAR) at the C-terminus of STIM [13, 32]. The C-terminus of Orai, which associates with STIM1 for SOCE function, exhibits variation among the three Orai homologues and results in differences in their affinity to STIM [28]. This STIM-Orai association ultimately drives Orai activation, causing Ca²⁺ influx into the cell [6].

Post-translational modification (PTM) of either STIM or Orai can modify their functions and their subsequent pathways. Post-translational modifications are the covalent attachment of a modifying group to residues in a protein [33]. The attachment of this group can alter the structure, localization, interactions, and overall function of the protein [33]. Here, we will discuss how STIM and Orai function can be modulated by PTMs, focusing primarily on phosphorylation, glycosylation, and oxidation by reactive oxygen species (ROS).

II. Phosphorylation

Phosphorylation is a reversible post-translational modification vital in both central regulatory, as well as a multitude of other signaling processes in cells [34]. Phosphorylation occurs through kinases, which are enzymes that catalyze the transfer of a phosphate from ATP to the molecule of interest. There are 3 classifications of kinases; tyrosine kinases, serine/threonine kinases, and dual specificity kinases, based on the residues they phosphorylate. In lower eukaryotes and plants there are no tyrosine kinases likely because they evolved more recently [34]. Phosphorylation of both STIM and Orai are mediated by protein tyrosine kinases and serine/threonine kinases, which can modulate SOCE in several ways [34]. Following phosphorylation, STIM and Orai can subsequently bind to downstream proteins [35], which in turn, leads to the creation of binding sites that modulate STIM-Orai interaction. While STIM1 phosphorylation is not required for SOCE activation, it increases the interaction between STIM and Orai, leading to increased SOCE [36]. Most kinases phosphorylate either serine and threonine, or tyrosine, but there are a few that can phosphorylate all of them [37]. Below, we will delve into the phosphorylation of different regulators of Ca²⁺ signaling, specifically proteins regulating SOCE, and how phosphorylation of these proteins regulates their functions.

Phosphorylation of STIM1

Phosphorylation of STIM1 by Tyrosine Kinases

STIM1 is a cell-surface phosphoprotein with many residues that undergo posttranslational phosphorylation [5, 19]. STIM1 contains tyrosine residues throughout different domains, several of which have been shown to be phosphorylated. Currently, it is believed that STIM1 tyrosine phosphorylation promotes STIM-Orai interaction, thereby increasing SOCE. STIM1 has six putative tyrosine kinase consensus sequences, two of which score high enough to be considered highly probable [38]. The two that scored high enough to be considered probable include Tyrosine 361 (Y361) within SOAR in the CC2 domain and Y316 within CC1 [39]. The Y361 residue can be phosphorylated by protein-rich kinase 2 (PYK2), modulating the interaction between STIM1 and Orai1[36]. This leads to increased recruitment of Orai1 to STIM1 puncta by promoting their interaction, thereby facilitating increased Ca²⁺ entry through Orai channels [38]. Note that phosphorylation of STIM1 is insufficient to cause STIM1-Orai1 interaction; this effect is dependent on ER Ca²⁺ depletion and the resulting reorganization of STIM1 into puncta (see Fig 2) [38]. Y316, is also predicted to affect SOCE, although to a somewhat lower degree and in the reverse direction. Hence, Y316 phosphorylation keeps STIM1in resting cells in a quiescent state. Further, Y316 phosphorylation regulates the interaction between STIM1 and SARAF [39], a modulatory protein that promotes disassembly of STIM-Orai complexes. There are several other known tyrosine residues in STIM1 that could serve as tyrosine kinase targets, however, whether or not this occurs and what functional impact their phosphorylation may have remains undetermined [38].

Figure 2: STIM1 Y361 phosphorylation contributes to STIM1 activation.

STIM1 phosphorylation on Y361 has been shown to contribute to the conformational change in STIM1 that facilitates Orai1 activation. Shown is a portion of cell spanning from the extracellular space to the ER; the ER has been depleted of Ca²⁺. On the left is depicted non-phosphorylated STIM1, in which some molecules interact with and activate Orai, but most remain in the resting conformation despite ER Ca²⁺ depletion. Once phosphorylated at Y361 (RIGHT), STIM1 takes on an active conformation, facilitating interaction with Orai and increased Ca²⁺ entry into the cell.

Phosphorylation of STIM1 by Serine/Threonine Kinases

Several potential serine phosphorylation sites have been discovered on STIM1, including S621, S575, and S608 [19]. These three residues are extracellular signal related kinase 1/2 (ERK1/2) target sites, and have been found to be phosphorylated during Ca²⁺ depletion in the ER, which plays a large role in SOCE activity [40]. In STIM1, the most common consensus sequences for serine/threonine kinases occur in sets of three and four residues. These include SXXE, TXXE, SXXD, TXXD SXK, TXK, TXR, and SXR [41]. Most of the known potential serine phosphorylation sites are concentrated in the ID and P/S domains; S621 and S608 are in the P/S domain [39]. Like tyrosine phosphorylation, phosphorylation of either S621 or S608 promotes interaction between STIM1 and Orai1, thereby increasing SOCE. Interestingly, S621 phosphorylation is lower in both OGT-overexpressing and siOGT-transfected cells than in wild-type cells, yet the phosphorylation of S575 and S608 were not affected by these modifications [42]. Decreasing the amount of phosphorylation at these serine residues reduces the likelihood of STIM-Orai interactions, which leads to lower SOCE.

Threonine 389 (T389) has been shown to be phosphorylated by protein kinase A (PKA) in a manner dependent upon the association between plasma membrane STIM1 and the scaffolding protein AKAP79. T389 is in the CC2 domain within SOAR, where it is well positioned to directly modulate STIM-Orai interaction [43]. Hence, reduced phosphorylation at this residue was shown to impair SOCE and promote the opening of arachidonate-regulated Ca²⁺-selective (ARC) channels, ER Ca²⁺-store independent channels created through an association between STIM1, Orai1, and Orai3 [43–45]. In conclusion, STIM1 phosphorylation of tyrosine, serine, and threonine residues is instrumental in STIM1 regulation with many potential targets available for future research.

Phosphorylation of STIM2

Given the high sequence similarity between STIM1 and STIM2, it seems likely that STIM2 is also a phosphoprotein, although which residues are phosphorylated has not been experimentally demonstrated [5]. Based on motif scans comparing kinase consensus sequences on STIM1 and STIM2, there are several matches [41]. Specifically, both tyrosine residues that are phosphorylated in STIM1 are also found in STIM2. Moreover, the sequences adjacent to STIM2_Y462 exhibit perfect sequence similarity with STIM1_Y361, making it highly likely that STIM2_Y462 would be phosphorylated in a manner like STIM1_Y361 [46, 47]. However, STIM2_T498 (homologous to STIM1_T389) [46, 47] and STIM2_Y417 (homologous to STIM1_Y316) [48] exhibit modest sequence variation in their surrounding residues. Although not experimentally tested, both STIM2T498 and STIM2Y417 would still be predicted to be targets of PKA [43] and a tyrosine kinase, respectively [49]. In contrast, the serine residue phosphorylation sites in STIM1 are found within the ‘variable’ region, which exhibits no sequence similarity to STIM2. However, STIM2 has a number of serine’s and threonine’s within its ‘variable’ region; whether or not these residues are phosphorylated and what the functional implications would be will need to be experimentally tested [48]. Finally, it should be noted that some of the sequence differences do affect function. For example, since STIM2 is not inserted into the PM [50], it is unlikely to contribute to ARC activity. Less is currently known about the phosphorylation of STIM2 than STIM1, but similarities in sequence allow assumptions about phosphorylation in STIM2 to be made that can be explored further in future research.

Phosphorylation of Orai

Orai1 is a tetra-spanning plasma membrane protein that has three intracellular regions containing putative PKC phosphorylation sites: the N terminus, C terminus, and an intracellular loop between transmembrane domain II and III [51]. PKC family is comprised of at least ten subunits which are divided into three subfamilies based on their primary structure. Since all PKC family members share common consensus sequences, they would be predicted to all phosphorylate Orai1, although only PKCβ has been directly shown to do so [51]. Hence, in vitro analysis revealed that only the N and C termini were phosphorylated by PKC, although phosphorylation was about 20 times higher in the N terminus [51]. It was further established that Ser-27 and Ser-30 are the key regulatory sites, with neutralizing mutations eliminating phosphorylation sensitivity [51]. In a subsequent study, our group showed that sustained phosphorylation of Orai1 at S27 and S30 leads to constitutive suppression of Orai1 activity in invasive melanoma [52]. Interestingly, the first 72 amino acids of Orai1 (consisting of S27 and S30) are not only unique to Orai1 but can be removed without attenuating Orai1 activation [53]. This is highly consistent with the idea that these residues serve a primarily regulatory role and somewhat suggestive that Orai1 is more prone to regulation than the other 2 isoforms although future investigations are needed to fully address this important question. Orai1 also has PKA phosphorylation sites, with Ser34 being the target of PKA-mediated phosphorylation in Orai1[54]. This direct phosphorylation mediates fast Ca²⁺-dependent inactivation (CDI) of Orai1 channels by calling more Orai1 channels into an inactive state. This decreased interaction between STIM1 and Orai1 leads to a lower SOCE [54].

III. Glycosylation

There is extensive evidence that members of the STIM and Orai families can be glycosylated. Glycosylation involves the covalent attachment of an oligosaccharide, also known as a glycan, to select resides of a target protein within the endoplasmic reticulum (ER)/Golgi apparatus, cytoplasm, or less commonly nucleus or mitochondria of eukaryotic cells [6, 55, 56] leading to differences in protein regulation [57], protein stability [58], and cell surface expression [59]. The two major forms of glycosylation are N-linked glycosylation and O-linked glycosylation [55]. In N-linked glycosylation, the oligosaccharide is attached to the amine of asparagine residues; typically occurring at Asn-X-Ser/Thr consensus sequences, where X can be any amino acid except proline [55]. O-linked glycosylation occurs when an oligosaccharide is attached to the hydroxyl group of serine or threonine. O-GlcNAcylation is a derivative of O-linked glycosylation that occurs through the attachment of ß-N-acetyl-glucosamine (O-GlcNAc) to serine or threonine of nuclear and cytoplasmic proteins and is the most common form of O-linked glycosylation in mammals [60]. Here we will discuss the mechanisms by which STIM1 and Orai are glycosylated and consider how these changes contribute to their function.

STIM1 N-linked Glycosylation

STIM1 and STIM2 can both be N-linked glycosylated within the sterile alpha motif (SAM) domain [5]. The N-linked glycosylation mechanism begins with two N-acetylglucosamines (GlcNAc) and dolichol pyrophosphate forming a linkage on the cytosolic side of the ER [61]. Dolichol is a long unsaturated chain that contains isoprene units and acts as a carrier in N-linked glycosylation [61]. The lipid linked glycan is then inverted to the luminal side of the ER and nine mannoses and three glucoses are added to form a more mature lipid linked glycan [61, 62]. Then oligosaccharyltransferase (OST) transfers the structure to the amide of asparagine residue that contains the consensus sequence -N-X-T/S [61] with the catalytic region of OST (STT3A/STT3B otherwise known as OSTC/TUSC3 respectively [63]) completing the transfer [64].

STIM1 has two N-linked glycosylation sites at Asn131 and Asn171. Analysis of the primary amino acid sequence of human STIM1 determined three consensus sites for N-linked glycosylation at N131, N171, and N658, however only N131 and N171 could only be modified in vivo [59]. Determination of these two sites was further verified through titration of STIM1 treated with Endoglycosidase H (EndoH) and truncated STIM1 that is terminated at the N658 site [59]. EndoH is an enzyme that cleaves glycan groups attached by N-linked glycosylation and is utilized to monitor N-linked glycosylation. When both variations of STIM1 were added to immunoblots, they migrated as a mixture of three isoforms: non-, mono-, and diglycosylated STIM1 species [59]. These results demonstrated STIM1 is N-linked glycosylated in vivo at both N131 and N171 [59]. In comparison to STIM1, STIM2 only has one site for N-linked glycosylation at the Asn232 site, which is analogous to the Asn131 site in STIM1 [58, 59]. STIM1 and STIM2 contain a single SAM (sterile α-motif) and an unpaired EF-hand in the extracellular region but diverge structurally in the C-terminal half of the cytoplasmic domain [5]. Although the interactions of STIM1 glycosylation have been investigated, the role of STIM2 glycosylation has not been determined.

STIM and Orai are primarily known for their role in the formation of the CRAC channel, however, they have also been shown to form ARC channels [44, 45]. ARC channels function independently of ER Ca²⁺ depletion and are activated by arachidonic acid production at the PM [65, 66]. While this activity has been demonstrated, the underlying mechanisms driving these distinct roles for STIM and Orai are somewhat unclear. One possibility is that PM-STIM1 contributes to ARC function; a fraction representing ~10% of STIM1 in the cell [18, 67–69]. This concept is based on several lines of evidence. First, tunicamycin, a non-selective blocker of N-linked glycosylation, blocks ARC activity [44]. In addition, although antibodies targeting the STIM1 N-terminus block CRAC channel function [68], in a subsequent study, inhibition of ARC channels was shown to occur with greater efficacy [44, 45]. Finally, the introduction of N131Q/N171Q (QQ) mutants led to a marked reduction in ARC activity [45]. However, while the concept of N-linked glycosylation at N131 and N171 contributing to STIM1 translocation to the PM is somewhat reasonable [59, 70], it has now been established that the STIM1-QQ mutant translocates to the PM with equal efficiency to STIM1_WT [71]. As such, the underlying mechanisms linking STIM1 glycosylation to ARC activity remains unclear.

Irrespective of its role in ARC activity, STIM1 glycosylation has been shown to contribute to CRAC channel activation [45, 72]. This was demonstrated through the smaller Ca²⁺ influx and smaller current density through CRAC channels facilitated by STIM1 glycosylation mutants relative to STIMWT in smooth muscle cells [45, 72]. This was established using both the STIM1_QQ mutant and a STIM1 N131D/N171Q (DQ) mutant, both of which eliminate N-linked glycosylation of STIM1 [71, 73]. Interestingly, while expression of STIM1_QQ led to decreased CRAC channel function, STIM1_DQ proved to be a gain of function mutation, driving increased current density and Ca²⁺ entry through store-operated channels [45, 71]. This was shown to be the result of increased translocation to the ER/PM junction resulting in a decrease in current latency through CRAC channels due at least in part to a partial destabilization of the EF-SAM domain [58, 71]. It was also observed that co-expression of STIM1_DQ with Orai1 led to a three-fold increase in the STIM1:Orai1 ratio relative to STIM1_WT: Orai1 [71]. In contrast, co-expression of STIM1_QQ with Orai1 decreased the STIM1:Orai1 ratio [71]. These observations reveal that N131 and N171 serve critical roles in control of STIM1 activation, likely due to the cross proximity of these residues to the EF and SAM domains. The fact that different mutations of these residues led to mutually opposing outcomes (see figure 3) reveals how functionally significant glycosylation of these residues would have to be, although how glycosylation would affect function is somewhat more difficult to assess given the profound impact of residue changes at these sites.

Figure 3: Glycosylation of STIM1 and N131 and N171 contributes to SOCE.

(A) STIM1WT is normally glycosylated at N131 and N171; under this condition, STIM1 responds to ER Ca²⁺ depletion with extension, oligomerization, and Orai1 activation. (B) Mutating both N131 and N171 to glutamine (QQ) decreases the STIM1/Orai1 ratio and subsequent SOCE, implicating glycosylation as a reaction that promotes STIM1 oligomerization and SOCE. (C) Mutating N131 and N171 to aspartic acid and glutamine (DQ), respectively, increase the STIM1:Orai1 ratio and subsequent SOCE, suggesting a key role for electrostatic interactions in STIM oligomerization.

Orai N-linked Glycosylation

Orai1 contains an N-linked glycosylation site at asparagine 223 (N223) that regulates CRAC channel function [74], while Orai2 and Orai3 do not contain any known N-linked glycosylation sites [26, 75]. Determining the functional role of N-linked glycosylation of Orai1 has been somewhat challenging due to cell type-specificity. Hence, in HEK293 cells, no differences in Orai1 localization or function were observed between Orai1-N223A and Orai1-WT [71], yet, in both Jurkat T cells and SCID fibroblasts, expression of Orai1-N223A led to enhanced SOCE, suggesting that glycosylation attenuates Orai1 function in these cell types [26, 76]. While the reasons for these differences are not fully established, it is likely a reflection of the fact that glycosylation patterns exhibit considerable cell type specificity due to differences in when the terminal sialic acid is added and which sialic acid-binding partners are expressed [75, 77]. Further, alterations to the glycosylation of Orai1 has been observed in aging, cancer, and immune disease [76, 77]. For instance, the enzyme β-Galactoside α-2,6-sialyltransferase 1 (ST6GAL1) has been implicated in cancer invasiveness and metastatic spread when elevated and has been shown to glycosylate Orai1, leading to reduced SOCE due to attachment of α-2,6-linked sialic acids [76]. In addition, sialic acid-binding immunoglobulin-type lectins (Siglecs) are glycan-binding proteins [76] with roles in mast cell and eosinophil-associated disease [76, 77]. Further, desulfation of Orai1 leads to an increase in Orai1-mediated Ca²⁺ entry and downregulation of Siglec-8 [76, 77]. Considered collectively, these data reveal Orai1 glycosylation as a fundamental contributor to cell type-specific differences in Orai1 function.

STIM1 O-linked Glycosylation

STIM1 function can be significantly altered by O-GlcNAc modification. The level of O-GlcNAcylation depends on glucose metabolism by the hexosamine biosynthesis pathway (HBP), which forms the substrate for O-GlcNAcylation [78, 79]. Extracellular glucose can be taken up by the cell through glucose transporter (GLUT) proteins [80]. After being phosphorylated and/or isomerized, the majority of this glucose is utilized in glycolysis, glycogen synthesis, or the pentose phosphate shunt, but about 2–5% fluxes through the HBP in typical somatic cells [81, 82]. Glutamine-fructose-6-phosphate aminotransferase 1 (GFAT1) is the first and the rate-limiting enzyme of the HBP, which converts fructose-6-phosphate (F6P) and glutamine into glucosamine6-phosphate (GlcN-6P) [83]. After subsequent acetylation and isomerization, newly formed N-acetylglucosamine-1-phosphate (GlcNAc-1P) goes through uridylation to yield UDP-GlcNAc, which is the substrate for O-GlcNAcylation [80]. O-GlcNAc addition to proteins is regulated by 2 opposing enzymes; O-GlcNAc transferase (OGT) attaches O-GlcNAc to proteins, while O-GlcNAcase (OGA) removes O-GlcNAc from protein to its unmodified state [84, 85]. It should be noted, OGT and OGA function in equilibrium, however, shifting the equilibrium dysregulates O-GlcNAcyIation, leading to pathology [86]. The effects of O-GlcNAcylation on STIM and Orai function will be discussed below.

O-GlcNAcylation of STIM1 has been shown to regulate its ability to facilitate SOCE [78]. As discussed in the introduction, ER Ca²⁺ depletion leads to the translocation of STIM1 to ER/PM junctions [87, 88], where it physically associates with Orai1, ultimately leading to its activation and Ca²⁺ entry [87]. Interesingly, STIM1 O-GlcNAcylation impairs its translocation to the ER/PM junction, which prevents STIM1-Orai1 association, thereby impairing SOCE [78, 89]. This to STIM1 translocation was shown by immunostaining of primary hepatocytes from genetically obese and lean wild-type (WT) mice treated with the irreversible SERCA pump inhibitor, thapsigargin (Tg), to induce ER Ca²⁺ depletion [89]. In WT hepatocytes from the lean mice, STIM1 is distributed throughout the ER at rest; Tg treatment led to marked STIM1 translocation to ER/PM junctions [89]. In contrast to hepatocytes from lean mice, in hepatocytes from obese mice treated with Tg, increased STIM1 signal intensity without translocation of STIM1 from the cytosol to the periphery of the cell was observed [89]. Notably, obese mice are known to have generally higher global O-GlcNAcylation levels [90], which has been specifically demonstrated for STIM1 [89]. Elevated O-GlcNAc levels can result from a higher flux through the HBP which have been demonstrated by an increase of HBP metabolites, induced by glucosamine [91, 92] or caused by hyperglycemia [93] and obesity [89], or through alterations to HBP enzymes, like in the inhibition of OGA or overexpression of OGT [89]. This is important because that indicates that obesity might lead to impaired STIM1 translocation and lower SOCE, which, may contribute to cardiovascular disease, diabetes (see Diabetes section) and, potentially, other pathological conditions.

Similar observations were made in neonatal cardiomyocytes treated with Tg [78]. STIM1 oligomerization and translocation to the ER-PM junction to form puncta structures are key features in STIM1-mediated SOCE. These features were seen in neonatal cardiomyocytes transfected with enhanced yellow fluorescent protein (eYFP)-tagged STIM1 and treated with EGTA and Tg to replicate Ca²⁺ depletion. However when O-GlcNAcyIation levels were increased with thiamet-G (TMG; highly-selective OGA inhibitor [94]) or glucosamine, there was a decrease in both the number of cells containing puncta and the average number of STIM1 puncta per cell [78]. Increased O-GlcNAcylation through metabolic stress in obesity or through the use of glucosamine or TMG has been shown to inhibit STIM1 oligomerization, translocation, and puncta formation, which are sequential steps in SOCE activation [78].

Since serine/threonine can be both phosphorylated and glycosylated [95], investigations into post-translational modifications of STIM1 can be somewhat complex. Initial interest in O-GlcNAcylation modification of STIM1 arose from the fact that STIM1 contains cytosolic regions rich in proline (P), glutamic acid (E), serine (S), and threonine (T), which has a high tendency to by modified by O-GlcNAcylation [6, 78]. Proteins that contain PEST sequences are often targeted for rapid degradation [96, 97]. However, O-GlcNAcylation of PEST sequences may slow or prevents proteins, like STIM1, from degradation [57]. STIM1 has a PEST sequence in the cytosolic C-terminal region [78, 96] and where O-GlcNAcylation occurs at both Ser621 and Thr626 [40]. Since Ser621 is also phosphorylated [98], this creates a scenario where the processes of phosphorylation and O-GlcNAcylation compete for the same residues, with functional consequences.

Computational prediction identified STIM1’s O-GlcNAcylation sites at Ser621 and Thr626 and these results were confirmed in HEK293 cells by co-immunoprecipitation assay. Further, since no O-GlcNAcylation occurred in an S621A/T626A STIM1 mutant, these may be the sole O-GlcNAcylation sites in STIM1 [40]. The fact that TMG treatment increased O-GlcNAcylation in STIM1_WT and STIM1_T626A but, not STIM1_S621A or STIM1_S621/T626A indicates that T626 is constitutively O-GlcNAcylated, while S621 O-GlcNAcylation is inducible [40]. Since S621 phosphorylation promotes STIM1 activation (see phosphorylation section), it is not surprising that O-GlcNAcylation at this site would lead to the opposite outcome. That seems to be the case, since TMG treatments resulting in S621 O-GlcNAcylation ultimately resulted in SOCE suppression. However, the relationship between Thr626 O-GlcNAcylation, Ser621 phosphorylation and SOCE is more complex.

To determine the relationship between STIM1 O-GlcNAcylation at T626 and phosphorylation of STIM1 at S621, STIM1-null HEK293 cells were transfected with STIM1_T626A exhibiting at least one of the phosphomimetic mutations S575E, S608E and/or S621E [40]. STIM1-KO cells expressing STIM1_S621E/T626A exhibited increased SOCE while no differences in SOCE were observed in cells expressing STIM1_T626A, STIM1_S575E/T626A or STIM1_S508E/T626A [40]. Therefore, this suggests that decreased STIM1 O-GlcNAcylation at T626 can lower SOCE activity by decreasing STIM1 phosphorylation at S621 [40]. Precisely how T626 O-GlcNAcylation controls S621 phosphorylation is not entirely clear, although it seems reasonable to speculate that it may increase access for modification of S621 through conformational changes.

Orai O-linked Glycosylation

The topic of Orai O-linked glycosylation has not been specifically explored. However, the established effects of STIM1 O-GlcNAcylation, the required STIM1-Orai1 association for SOCE and the cytosolic domains of Orai1, may point to the possibility of Orai1 being regulated by O-linked glycosylation [78].

O-GlcNAc in Diabetes

Diabetes is characterized by high blood glucose concentrations (hyperglycemia) caused by either the inability to produce insulin due to destruction of β-cells in the pancreas islet to (Type I) or the inability to use insulin due to resistance of insulin receptors (Type II) [99, 100]. As previously mentioned, the substrate for O-GlcNAcylation (UDP-GlcNAc) is synthesized from glucose in the HBP and in cases of high concentrations of glucose, as seen in diabetes, HBP is upregulated [79].

Increased O-GlcNAc levels have been associated with cancer, neurogenerative diseases, and diabetes. The increase of O-GlcNAc content and elevated HBP flux observed in diabetes has an unfavorable effect on many cells and tissues, particularly cardiomyocytes [78]. For instance in the cardiomyocytes of type 2 diabetic mice, phenylephrine and angiotensin II induced hypertrophic signaling was blunted due to elevated HBP flux and increased protein O-GlcNAc levels [78]. Hypertrophic signaling is associated with diabetic cardiomyopathy defined by myocardial dysfunction in the absence of coronary artery disease [93]. Although increases in O-GlcNAc can decrease hypertrophic signaling, it is not a long-term benefit since O-GlcNAc levels tends to be an indicator of an endogenous stress response [78]. Additionally, inhibition of HBP partially reversed the effect of blunted hypertrophic signaling [93].

Since diabetes increases the HBP flux and O-GlcNAc levels, this has implications for SOCE. The increased O-GlcNAc levels resulting from diabetes has been implicated in the inhibition of SOCE in smooth muscle cells, cardiomyocytes, macrophages, and even the platelets of diabetic patients [93, 101, 102] In cardiomyocytes, diabetes decreased SOCE which was induced by thapsigargin or angiotensin II. This attenuation of SOCE by hyperglycemia could be prevented using azaserine, which is an inhibitor of the HBP.

IV. Acylation

A recent study revealed acylation as another post-translation modification of Orai. Acylation is a lipidation modification that attaches acyl groups to proteins and in S-acylation the attachment of acyl groups occurs at cysteine [103, 104]. S-acylation typically occurs at the cytosolic side of membranes, which is consistent with Orai1’s single S-acylation site at cysteine 143 (C143) in the cytosolic TM2/TM3 loop [103]. Orai1 can be reversibly S-acylated using DHHC21, a protein acyltransferase with a common DHHC (Asp-His-His-Cys) motif [105]. When ER Ca²⁺ store depletion occurs, DHHC21 will rapidly S-acylate Orai1, which results in Orai1 interaction with STIM1 at the plasma membrane and CRAC channel activation [103]. A mutation of the S-acylation site to serine (C143S) in Jurkat T cells results in decreased SOCE due to the reduced STIM1-Orai interaction after Ca²⁺ store depletion when compared to WT Orai1 [103]. Further, S-acylated Orai1 has been shown to facilitate targeting to lipid rafts and increase the efficiency of Ca²⁺ signals at lipid rafts during the formation of the immunological synapse [104] and likely many other contexts.

Myristoylation is a specific acetylation process that can affect Orai1 function. Myristoylation is a lipidation modification that attaches a myristoyl group to the N-terminus of glycine using N-myristoyltransferase (NMT) [106]. The myristoyl group is derived from myristic acid, a fatty acid with a 14-carbon backbone [106]. Although STIM1 or Orai1 are not known to be myristoylated, myristoylation of Golli proteins have been shown to interact with STIM1 and alter SOCE. Golli proteins are encoded by the myelin basic protein (MBP) gene and is most expressed in immune tissues with the isoform Golli-BG21 being the most predominant [107]. Golli-BG21 is involved in Ca²⁺ regulation by inhibiting SOCE [107]. Golli-BG21 does contain a myristoylation site that permits PM association and is involved in the attenuation of SOCE [107–110]. In the formation of Golli-BG21 myristoylation mutation (MYR-BG21), the glycine at position 2 was mutated to alanine [107]. In Jurkat T-cell transfected with WT-BG21 or MYR-BG21, MYR-BG21 was absent from the PM using immunoblots and demonstrated greater Ca²⁺ in comparison to its WT counterpart [107]. Therefore, myristoylated Golli-BG21 allows PM association and its ability to act as a negative regulator of SOCE. The process of Golli-BG21 inhibition of SOCE at the PM involves its interaction with STIM1 at its C-terminus after Ca²⁺ store depletion [108–110]. The overexpression of STIM1 can reverse Golli-BG21 inhibitory effect on SOCE even though the expression of Golli-BG21 remains constant [110]. Although the mechanism between the interaction of Golli-BG21 and STIM1 is still unknown, since Golli-BG21 and Orai1 can both interact at the C-terminus of STIM1, Golli-BG21 may compete with Orai1 to interact with STIM1 and reduce SOCE [108].

V. Redox Modulation of STIM and Orai

Reactive oxygen species (ROS) and reactive nitrogen species (RNS) include both radical and nonradical reactive species obtained from the metabolism of molecular oxygen and nitrogen respectively [111]. ROS and RNS include many different substances, with the most biologically relevant molecules being superoxide (O₂-), hydrogen peroxide (H₂O₂) and nitric oxide (NO) [111]. ROS and RNS participate in the regulation of cellular processes like proliferation, metabolism, and differentiation [112]. However, overproduction of ROS and RNS leads to damage of nucleic acids, proteins, and lipids [112], which will be discussed in further detail below.

Oxidative damage IS prevented by the presence of ROS/RNS scavengers, the glutathione (GSH) cycle, and protective antioxidant enzymes such as superoxide dismutase or catalase. For instance, the protective enzyme superoxide dismutase converts the highly unstable superoxide into the more stable hydrogen peroxide (a radical-derived, non-radical reactive species) [111, 113]. Glutathione is an abundant antioxidant capable of neutralizing both ROS and RNS, functioning as a cellular redox buffer [114]. Hence, when GSH encounters either ROS or RNS, it absorbs the reactive electron; the resultant reactive GSH forms a disulfide bridge with another glutathione molecule, producing the dimer glutathione disulfide (GSSG) [114]. With the help of glutathione reductase, glutathione is recycled by the reduction of glutathione disulfide back to glutathione, facilitating sustained maintenance of redox homeostasis [111, 114].

Many exogenous and endogenous sources can produce ROS. Exogenous sources include environmental sources like heat, radiation, and chemical substances [113]. However, endogenous sources contribute to the majority of ROS production and involve oxidative enzymes and metabolic pathways like the electron transport chain [113]. In the electron transport chain located in the mitochondria, electrons are transferred through each of the four protein complexes to eventually reduce oxygen to form H₂O [115]. However, some electrons can be diverted from this pathway, particularly in complex II, and participate in the single-electron reduction of oxygen to produce a superoxide [115, 116]. Another endogenous source of ROS is from oxidative enzymes such as NADPH oxidases (NOX) [113]. NOX is primarily located at the plasma membrane, although in phagocytes it is also found on phagosomes [113]. NOX generates ROS by transferring electrons from cytosolic NADPH to free oxygen molecules, forming superoxide [113]. 5-Lipooxygenases (LOX), cyclooxygenases (COX), xanthine oxidases are additional oxidative enzymes that contribute to ROS production [113].

Nitric oxide (NO) is one of the main radical RNS and is formed from the metabolism of L-arginine [117]. Nitric oxide synthases (NOS) are utilized to oxidize the guanidine nitrogen of L-arginine into L-citrulline and nitric oxide [117].

Although ROS and RNS can cause oxidative damage at high concentrations, at lower concentration they can function as second messengers under physiological conditions through post-translational protein modifications that regulate gene expression, cellular signaling, and diverse homeostatic functions [113, 118]. ROS can pass through membranes using passive diffusion or aquaporins which are H₂O₂-permeable [118]. Upon entry into the cytosol, ROS species are short-lived due to the high availability of protective antioxidants enzyme superoxide dismutase, catalase, or through glutathione reductase system [113]. However, in cases of reduced antioxidant enzymes or increased production of ROS or RNS, oxidative stress can cause post-translational modification on proteins [111]. In conclusion, ROS and RNS have a surprising plethora of functions within cells including modulation of proteins which will be paramount in the following discussion.

Due to the high reactivity of cysteine residues within proteins, cysteine residues act as the main targets of oxidation [118]. S-sulfenylation, S-nitrosylation, and S-glutathionylation are examples of cysteine oxidation products that are involved in redox-dependent regulation of proteins [113]. In S-sulfenylation, peroxide oxidizes the thiolate groups (S-) on cysteine reversibly into sulfenic acid (R-SOH) or irreversibly into sulfinic or sulfonic acid [119]. S-sulfenylation acts as a fleeting molecular switch that can shift between different stable states depending on the local environment, allowing for further chemical modifications, such as S-glutathionylation and S-nitrosylation [113, 118]. After the formation of sulfenic acid by S-sulfenylation, glutathione can be reversibly added in the process known as S-glutathionylation [113]. S-glutathionylation acts as an antioxidant defense system that helps to prevent permanent oxidative damage of proteins like STIM1 by adding a tripeptide known as glutathione, composed of γ-glutamic acid, cysteine, and glycine, to reactive cysteines that are susceptible to oxidative stress [120]. The process of S-glutathionylation is reversible as glutathione-S-transferase catalyze the addition of glutathione, while glutaredoxin removes the glutathione and requires free glutathione as a cofactor [113, 120]. In a similar process called S-nitrosylation, a nitric oxide can be reversibly added onto reduced cysteine residues to form S-nitrosocysteines [121]. As discussed further below, STIM and Orai are both proteins that experience redox modifications that regulate their function and subsequently regulate SOCE.

ROS Modification of STIM1

Human STIM1 contains five cysteine residues, including one in the signal peptide (C4), one in the transmembrane domain (C227), one in the C-terminal STIM-ORAI activating region, known as SOAR (C437), and two in the N-terminal ER luminal domain (C49, C56) [122]. C4 is not a good candidate for redox modulation due to being localized in the signal peptide and is cleaved before being translocated to the ER [122]. Likewise, C227 is not a good candidate due to being located close to the cytoplasmic exit of the transmembrane domain [123]. However, C49 and C56 are two highly conserved cysteine residues located in the N-terminal region that are potentially susceptible to ROS or S-glutathionylation modulation [122]. Due to their high conservation and in proximity to the Ca²⁺-sensing EF-hand, the potential for C49 and C56 to serve as redox sensors has been assessed [124]. Several strategies were used to induce oxidative stress including exogenous H₂O₂, lipopolysaccharides (LPS)-induced stress and buthionine sulfoximine (BSO), which blocks GSH recycling. [124]. Using these strategies, it was established that STIM1 is S-glutathionylated on C56, but not C49 under conditions of oxidative stress. This led to a decrease in the affinity of STIM1 for Ca²⁺, ultimately resulting in STIM1 oligomerization and Orai1 activation in the absence of ER Ca²⁺ depletion [124].

Interestingly, co-transfection of STIM1-null mouse embryonic fibroblasts with Orai1 and either STIM1 WT or a double mutant (C49A/56A) suggested that C49 and C56 are required for STIM1 function since SOCE was only detected in STIM1 WT expressing cells [125]. This was somewhat unexpected since STIM1 (C56A) was found to be constitutively active [124]. Whether these differences reflect cell type differences or the importance of C49 is not entirely clear. Irrespective, both studies support the concept that STIM1 may function not only as a Ca²⁺ sensor but also as a redox sensor [124, 125].

STIM1 cysteine residues at C49 and C56 have also been shown to interact with the ER oxidoreductase ERp57 to modulate SOCE in mouse embryonic fibroblasts [120]. Notably, SDS-PAGE in non-reducing and reducing conditions revealed that C49 and C56 form an ERp57-induced disulfide bridge in STIM1. In addition, ERp57 binding to STIM1 decreases puncta formation and SOCE function, while the deletion of ERp57 results in the opposite effect [120, 125]. Considered collectively, these data could indicate that the formation of the disulfide bond between C49 and C56 by ERp57 prevents the S-glutathionylation of C56, thereby decreasing SOCE function.

Comparable to the role that C56 serves as a sensor for oxidative stress and can lead to the activation of SOCE, C437 may be involved in the opening of CRAC channels. C437 is a cytosolic cysteine within the CAD/SOAR region and was mutated to glycine (C437G) to understand it role in inducing SOCE [17]. STIM1_C437G migrates to the PM and co-localizes with Orai1 upon store-depletion at a similar rate to WT-STIM1 [122, 126]. However, SOCE developed at a 4-fold lower rate in STIM1_C437G-expressing cells than WT-STIM1 expressing cells, indicating C437 may not be involved in STIM1-Orai1 coupling but could be critical for driving the conformational change in Orai1 that precedes channel opening [122, 126]. While the extent to which C437-dependent effects on SOCE are ROS have not been explored, given the sensitivity of cysteines to redox modulation, this possibility cannot be discounted.

ROS Modulation of STIM2

STIM2 contains 15 cysteine residues: there are 4 cysteines at C140, C147, C318, and C528 corresponding to cysteines found within STIM1 and 11 unique cysteines within the cytosolic domain [127]. Although ROS modulation of STIM2 has not been extensively investigated, the additional cysteines in STIM2 suggest a greater potential for redox modulation [122]. Like STIM1, STIM2 acts Ca²⁺ sensor but its EF-hand domain has a lower affinity for Ca²⁺, allowing it to respond to very modest changes in ER Ca²⁺ concentration [14]. Like STIM1, STIM2 oligomerizes and associates with Orai upon activation, although it is a much less effective activator of CRAC current [15, 16].

To understand how oxidative stress affects STIM2-mediated SOCE, melanoma cells were utilized, which contain STIM2 as their predominant STIM isoform [52, 127]. WM3734 Melanoma cells treated with H2O2 demonstrated reduce SOCE in STIM2 silenced with siRNA and in STIM2 non-silenced cells. When compared to their respective controls, STIM2 significantly decreased the redox sensitive of SOCE in melanoma cells. These results were also verified in a second melanoma cell line (1205Lu) and human CD4⁺ T cells.

OxlCAT redox proteomics and additional assays identified C389 and C400, both STIM2-specific cysteines localized in the CC1 domain, as being potential oxidative sensors [127, 128]. However, cross-species alignment best suggests C400 as a more notable oxidative regulator due to being more conserved residue than C389 among other species [127]. To understand the functional role C389 and C400 contributes to the oxidative sensitivity of SOCE, HEK293 cells (HEKO1) were overexpressed with Orai1 and with no cysteines. STIM2_C389V, STIM2_C400V, or STIM2_WT were overexpressed in in HEKO1 cells treated with H2O2 [127]. STIM2_C389V demonstrated partially reduced H₂O₂-induced SOCE inhibition, while the STIM2_C400V mutation almost completely eliminated H₂O₂-induced SOCE inhibition [127]. Therefore, unlike STIM1, STIM2 responds to oxidation with SOCE inhibition, with C400 being required as a redox sensor and C389 functioning contributing to oxidation-induced STIM2 activation, but not being absolutely required [127]. Additionally, FRET and fluorescence lifetime imaging microscopy (FLIM) experiments establish that oxidation of C400 interferes with STIM2 activation and oligomerization [127], likely by stabilizing CC1 in the inactive state.

In the process of STIM activation, when either STIM1 or STIM2 cluster at the ER-PM junction, STIM associates with negatively charged lipids within the inner leaflet of the PM, which promotes its interaction with Orai1 [69, 129]. Although STIM2 has lower Ca²⁺ affinity within the EF-hand domain, its C-terminal K-rich domain (see Fig 1A), has a higher affinity for negatively charged lipids in the PM, particularly PIP₂ [69]. This increased affinity for PIP₂ is primarily due to dimerization of the STIM2 K-rich domain, allowing it to form a amphipathic α-helix, which is not seen during STIM1 activation [130]. Further, oxidation of STIM2_C725, located near the K-rich domain, is required for dimerization to occur [130]. Although established in liposomes, the extent to which this occurs during SOCE in intact cells has yet to be determined. Nevertheless, these observations reveal yet another type of redox modulation of STIM2.

RNS Modification of STIM1

While the roles of C49 and C56 in STIM1 as oxidative sensors was described above, they have also been shown to serve as nitrosative sensors [121]. C49 and C56 are susceptible to modulation of STIM1 by RNS in a process called S-nitrosylation. As previously mentioned, S-glutathionylation of STIM1 destabilized the EF-hand domain, decreasing the affinity of the EF hand for Ca²⁺, thereby causing STIM1 activation and Orai1 activation [121, 124]. In contrast, S-nitrosylation of C49 and C56, facilitates an interaction between the STIM1 24–57 region with the EFSAM core [121]. This stabilized the STIM1 EF-hand domain without interfering with Ca²⁺ binding [121]. Further, SOCE measurements in the presence or absence of the NO donor S-nitrosoglutathione (GNSO), revealed that this structural shift leads to SOCE inhibition [121]. Finally, it was shown that STIM1_C49S/C56S facilitated decreased SOCE in the presence or absence of GNSO, suggesting that C49 and C56 mediated S-nitrosylation-mediated SOCE inhibition [121].

Similar observations were made using mouse neonatal cardiomyocytes which contains neuronal nitric oxide synthase located in the sarcoplasmic reticulum. NMR studies demonstrated that when NO targets C49 and C56, it results in decreased mobility of STIM1 24–57 region [131]. Due to the increased stability of STIM1, oligomerization and SOCE was suppressed in the presence of Ca²⁺ depletion [131]. These findings indicate that nitric oxide production is associated with S-nitrosylation of STIM1 and serves as an endogenous suppressor of SOCE.

RNS Modification of STIM2

Similar to STIM1, homologous cysteines in STIM2 (C150 and C157) have been shown to be nitrosylated [132]. Further, STIM2 has an additional cysteine residue (C112) that is also capable of thiol modification under nitrosative stress [132]. S-nitrosylation of STIM2 was found to significantly stabilize the EF-SAM domain thermodynamically due to interactions of the N-terminal region of STIM2 with a single face of the EF-SAM. The stabilization of the EF-SAM domain prevents STIM2 activation, through a similar mechanism to its effect on STIM1 [132]. Further, GNSO inhibits STIM2-mediated SOCE in HEK293 cells overexpressing STIM2_WT whereas GNSO had no effect on STIM2_{C112C150C157-mediated} SOCE.

Interestingly, increased expression of STIM2 has been associated with many diseases like pulmonary arterial hypertension [132, 133]. Patients with pulmonary arterial hypertension exhibit upregulation of STIM2 in their pulmonary arterial smooth muscle cells in comparison to STIM1 [134]. Since STIM2 is highly sensitive to S-nitrosylation in comparison to STIM1 due the presence of additional cytosolic cysteines, this may represent a potential therapeutic option [127, 132, 133]. The initial study proposed that in platelets, NO donors was able to indirectly inhibit SOCE through activation of cGMP dependent protein kinase and possible phosphorylation of STIM2 [132, 134]. However, since STIM2 can be directly affected by NO donors, S-nitrosylation most likely stabilized the STIM2 EF-SAM domain and prevented STIM2 activation and subsequent SOCE function [132].

ROS Modulation of Orai (Orai1, Orai2, Orai3)

Like STIM proteins, Orai1 and Orai2 are also susceptible to redox modulation at the homolog site C195 that results in the inhibition of SOCE. Orai1 contains three cysteines sites that are ROS targets by H₂O₂ [135]. Orai1 contains cysteines at C126, C143, and C195, which are all susceptible to ROS modulation. Orai2 also contains three cysteines at homologous sites, while Orai3 does not contain the third homologous site at C195, but two additional cysteines between T3 and T4 [120].

Interestingly, due to the lack of the homologous C195 site within Orai3, it results in redox resistance which indicates the importance of C195 in redox regulation in SOCE as a redox sensor [135]. The redox significance of the C195 was analyzed by forming a gain-of-function mutation in Orai3 by mutating the homologous C195 site in Orai3 (G170) to cysteine, forming Orai3_G170C [135]. The mutation Orai3_G170C created redox sensitivity in Orai3, but it demonstrated similar small currents to G170 in response to H₂O₂, indicating Orai3 is redox resistant [135]. Although ROS inactivates CRAC channels that lack an Orai3 subunit, when an Orai3 subunit is present, the Orai3/Orai1 ratio determines the overall redox sensitivity of the CRAC channel [136]. The presence of a single Orai3 subunit can form redox resistance in Orai1/Orai3 heteromeric channels.

In Orai1, C195 oxidation by H₂O₂ results in the inhibition of SOCE in vascular smooth muscle cells, HEK293 cells, Jurkat T cells, and CD4⁺ T cells in both Orai1 and Orai2 [135, 137]. Using FRET microscopy, the effect of oxidation on STIM1-Orai1 interaction and Orai1-Orai1 subunit interaction was determined [138]. Interestingly. H₂O₂ pretreatment resulted in a 46% decrease in Orai1-Orai1 subunit interaction, while STIM1-Orai1 interaction was increased [138]. The fact that Orai1 subunit interactions were decreased presumably explains the loss of function observed when C195 oxidized; that STIM1 and Orai1 can still interact is likely an epiphenomen that would not lead to channel activation in the absence of fully intact pore forming units.

Redox Summary

As outlined above, overproduction of ROS/RNS can have multifaceted, complex and, in some cases, mutually contradictory effects upon STIM/Orai conformation and function. While it is often tempting to try to make blanket statements about what effect redox modulation may ultimately have on Ca²⁺ entry, the work described above suggestions that the localization and identity of the species involved can lead to very different outcomes. Given the pleiotropic effects of Ca²⁺ signals on cellular function, whether or not redox changes activate or inhibit Orai activation may have a profound impact on the physiological/pathophysiological outcome. As such, future investigations defining these relationships in realistic scenarios may provide valuable mechanistic insights redox-mediated control of cell function.

Conclusions

Here we have provided a summary of what is currently known about the post-translational modifications of members of the STIM and Orai families. While the function of STIM and Orai is primarily controlled by ER Ca²⁺ content (by definition) it is clear that both the degree of ER Ca²⁺ dependence and the extent to which the proteins respond is subject to considerable modification through PTMs and/or interactions with accessory proteins. As highly conserved proteins found in virtually every cell of metazoan species, it is tempting to imagine that store-operated Ca²⁺ entry would occur in much the same way in all cell types. However, through PTMs such as described here, it is likely that physiological SOCE may exhibit considerable cell type variability in scale and consequence, particularly when measured in response to physiological agonists. Now that the common mechanisms of SOCE are largely defined, future investigations focused on demonstrating and explaining cell type and context-dependent differences in this signaling pathway may be highly revealing. Further, considering that some of these differences occur in a disease context, targeting these modifications may have greater therapeutic potential than less selective strategies that might affect SOCE in all cell types.

Figure 4: Species-dependent control of STIM1 activation by redox modulation.

(A) ROS overproduction can cause S-glutathionylation of STIM1 at C56 within the ER lumen, resulting in destabilization of the EF-hand domain and decreased Ca²⁺ affinity of STIM1. The destabilization of STIM1 promotes activation, oligomerization, STIM1-Orai1 coupling, and subsequent SOCE function. (B) S-nitrosylation of STIM1 at C49 and C56 due to overproduction of nitric oxide within the ER lumen results in stabilization of the EF-hand domain and increased Ca²⁺ affinity of STIM1. The increased stabilization of STIM1, prevents STIM1 activation and inhibits SOCE function.