Interplay between ER Ca2+ Binding Proteins, STIM1 and STIM2, Is Required for Store-Operated Ca2+ Entry

Store-operated calcium entry (SOCE), a fundamentally important homeostatic and Ca2+ signaling pathway in many types of cells, is activated by the direct interaction of stromal interaction molecule 1 (STIM1), an endoplasmic reticulum (ER) Ca2+-binding protein, with Ca2+-selective Orai1 channels localized in the plasma membrane. While much is known about the regulation of SOCE by STIM1, the role of stromal interaction molecule 2 (STIM2) in SOCE remains incompletely understood. Here, using clustered regularly interspaced short palindromic repeats -CRISPR associated protein 9 (CRISPR-Cas9) genomic editing and molecular imaging, we investigated the function of STIM2 in NIH 3T3 fibroblast and αT3 cell SOCE. We found that deletion of Stim2 expression reduced SOCE by more than 90% in NIH 3T3 cells. STIM1 expression levels were unaffected in the Stim2 null cells. However, quantitative confocal fluorescence imaging demonstrated that in the absence of Stim2 expression, STIM1 did not translocate or form punctae in plasma membrane-associated ER membrane (PAM) junctions following ER Ca2+ store depletion. Fluorescence resonance energy transfer (FRET) imaging of intact, living cells revealed that the formation of STIM1 and Orai1 complexes in PAM nanodomains was significantly reduced in the Stim2 knockout cells. Our findings indicate that STIM2 plays an essential role in regulating SOCE in NIH 3T3 and αT3 cells and suggests that dynamic interplay between STIM1 and STIM2 induced by ER Ca2+ store discharge is necessary for STIM1 translocation, its interaction with Orai1, and activation of SOCE.


Introduction
Dynamic changes in the concentration of free calcium ions within the cytosol ([Ca 2+ ] c ) are universal intracellular signals that act over a wide temporal and spatial range to control many cell functions. Ca 2+ signals are generated by the mobilization of calcium into the cytosol via influx through Ca 2+ -permeable ion channels in the plasma membrane (PM) and following release of Ca 2+ from sites of intracellular sequestration.
Store-operated Ca 2+ entry (SOCE), a specific type of Ca 2+ influx mechanism, is an important Ca 2+ signaling pathway in both excitable and non-excitable cells that is activated following release of Ca 2+ sequestered within lumen of the endoplasmic reticulum (ER). SOCE provides local and global Ca 2+ signals that regulate Ca 2+ -dependent biochemical events and is a source of Ca 2+ for refilling depleted ER Ca 2+ stores following cellular stimulation. The SOCE mechanism, also called capacitative Ca 2+ entry, was first proposed by James Putney in 1986, but the molecular basis of SOCE was not discovered until 2005 and 2006 [1]. In siRNA screens, independent laboratories identified stromal interaction molecule 1 (STIM1) as an ER Ca 2+ sensor responsible for the activation of SOCE [2,3]. Shortly after this discovery, genome-wide RNAi screens revealed that Orai1 was the ion-conducting pore subunit of store-operated channels (SOCs) [4,5].
Following the identification of STIM1 and Orai1 as the pivotal regulatory components of SOCE, numerous studies of a variety of cell types have led to the formulation of a consensus model that describes the mechanisms underlying the activation and inactivation of SOCE. In unstimulated, resting cells, STIM1 is diffusely distributed throughout the ER membrane, where it exists either as a monomer or as dimers [6][7][8][9]. Subsequent to exposure of cells to stimuli that cause release of ER Ca 2+ , the reduction in ER Ca 2+ concentration ([Ca 2+ ] ER ) leads to the dissociation of Ca 2+ from the Ca 2+ -binding helix-loop-helix structural domain (EF-hand motif) of STIM1, which consequently induces a conformational change and oligomerization of STIM1. This is followed by a rapid translocation and accumulation of STIM1 in discrete multi-protein clusters or punctae in plasma membrane-associated ER (PAM) nanodomains, subcellular regions in which the PM and ER membranes are in close apposition and functionally interconnected [3,10]. Current evidence suggests that within PAM sites, STIM1 binds to and activates Orai1 Ca 2+ channels [3,6,11,12] via an interaction between a cytosolic motif of STIM1 known as the STIM1 Orai activating region (SOAR) and a carboxy terminus α-helical region of Orai1 [13][14][15].
STIM2, an ER single-pass transmembrane protein and homologue of STIM1, has also beenidentified, but relatively little is known about the roles of STIM2 in SOCE and other cellular functions. STIM1 and STIM2 are largely conserved and share high amino acid sequence homology. They both have a Ca 2+ -binding EF-hand domain, a sterile α motif (SAM) domain, three sequential coiled coil (CC) domains containing the SOAR, and a carboxy terminus lysine-rich domain [16]. It is of note, however, that the STIM proteins diverge considerably in the carboxy terminus half of the cytoplasmic domain, suggesting potential functional differences. Additionally, the Ca 2+ binding affinity and activation kinetics of STIM2 differ from that of STIM1 [17][18][19]. STIM2 exhibits a lower Ca 2+ affinity than STIM1. In a Ca 2+ -unbound state, the STIM2 EF-SAM domain is much more stable as a monomer; unlike STIM1, STIM2 does not readily aggregate or form puncta. Furthermore, a single amino acid difference in the STIM2 SOAR domain makes it a poor activator of Orai1 compared to STIM1 [20]. These differences in critical functional domains may contribute to alternate functions of the STIM proteins. The literature suggests that STIM2 functions as a homeostatic regulator of cytosolic and ER Ca 2+ content, while it has only a minor contribution to SOCE in some cell types and none in others [2,3,17,21]. STIM2 has been shown to be active subsequent to small amplitude decreases in [Ca 2+ ] ER [17,[21][22][23][24]. In sharp contrast, stable overexpression of STIM2 in HEK293 cells inhibits SOCE [25].
In order to provide a greater understanding of how STIM2 participates in Ca 2+ homeostasis and signaling, we used clustered regularly interspaced short palindromic repeats (CRISPR)-Cas9 genomic editing and molecular imaging to determine the roles of STIM2 in Ca 2+ homeostasis and SOCE in NIH 3T3 cells, a fibroblast cell line derived from mice, and αT3 cells, an immortalized gonadotroph cell line from mice. Our hypothesis was that STIM2 would be important for both Ca 2+ homeostasis and SOCE. Our studies confirmed that STIM2 participates in the regulation of basal [Ca 2+ ] c in unstimulated NIH 3T3 and αT3 cells, and is necessary for activation of SOCE following depletion of ER Ca 2+ stores. We also found that STIM2 was required for STIM1 translocation and puncta formation, as well as essential for the interaction of STIM1 with Orai1 in NIH 3T3 cells. Taken together, our findings suggest STIM2 recruits STIM1 and stabilizes it in PAM nanodomains to facilitate interaction with and activation of Orai1 channels.
In most types of cells, Orai1, a highly selective Ca 2+ channel, is considered to be the primary SOC channel activated by STIM1 in response to ER Ca 2+ store depletion [30]. In addition to the Orai proteins, members of the transient receptor potential canonical (TRPC) ion channel family have also been shown to be involved in Ca 2+ entry in response to Ca 2+ -store depletion [31,32]. Unlike Orai channels, TRPC channels are Ca 2+ -permeable non-selective cation channels [33,34]. Ionic substitution in extracellular bathing solutions was used to characterize the Ca 2+ -selectivity of the channels mediating SOCE in the NIH 3T3 cells. Replacement of external Na + with an equimolar concentration of N-methyl-D-glucamine (NMG), a cation that enters and blocks the ion-conducting pore of non-selective channels, but not Ca 2+ selective channels, had no effect on SOCE. This suggests that SOCE is mediated by a channel that exhibits Ca 2+ selectivity in the NIH 3T3 fibroblasts ( Figure 2). Our findings suggest that Orai1 is responsible for the SOCE in NIH 3T3 fibroblasts. To further test the hypothesis that Orai1 specifically is responsible for the SOCE response we were measuring, a dominant-negative Orai1 mutant, Orai1 E106A, was transiently expressed in the cells. In cells expressing Orai1 E106A, SOCE was nearly completely abolished, consistent with Orai1 playing a predominant role in SOCE ( Figure 3). Since we were able to establish that SOCE occurs through Orai1 and can be attenuated by several SOCE inhibitors, we next sought to identify the role STIM2 plays in this process. Figure 1. Store-operated Ca 2+ entry in NIH 3T3 cells. (A) Store-operated Ca 2+ entry (SOCE) in fibroblasts following passive endoplasmic reticulum (ER) Ca 2+ -store depletion with cyclopiazonic acid (CPA), an inhibitor of sarco/endoplasmic reticulum Ca 2+ -ATPase (SERCA). Cells grown on glass coverslips were loaded with Fura-2 and perifused with test solutions at 37 °C. Cells were initially bathed in standard extracellular solution (SES) containing 1 mM Ca 2+ to establish a baseline. The bath solution was changed to a Ca 2+ -free SES (open bar; Ca 2+ -free), followed by Ca 2+ -free SES with 20 μM CPA to discharge and prevent refilling of the intracellular Ca 2+ stores. SOCE was triggered by the subsequent application of SES with 1 mM Ca 2+ (open bar; +Ca 2+ ) in the continued presence of CPA; the rise of cytosolic [Ca 2+ ] indicative of SOCE in single cells was measured using quantitative fluorescence microscopy and is expressed as the increase in the ratio of Fura-2 FI340/FI380. Data are expressed as the mean ± SEM (n = 67 cells). (B) Pharmacological analysis of SOCE. Cells were loaded with Fura-2 in SES and Ca 2+ was measured with the FlexStation 96-well plate reader. The cells were incubated in a Ca 2+ -free SES containing CPA (20 μM) and the SOCE inhibitor or vehicle control. Application of GSK-7975A (GSK; 50 μM), SKF96365 hydrochloride (SKF; 10 μM), ML-9 (100 μM), or NAGly (30 μM) before the reintroduction of extracellular Ca 2+ (open bar; +Ca 2+ ) caused a significant reduction in the (C) normalized peak amplitude (Max FI335/FI375) and (D) area under the curve (AUC60) of the store-operated Ca 2+ response. Graph data in (B) are plotted as the time-dependent change in the mean ± SEM of the fold change in the ratio of Relative FI335/FI375, averaged from 12 or more wells for each inhibitor from at least three independent experiments. In the box and whisker plots, the center solid line marks the median, small open square within the box depicts the mean, the ends of the box are the 25th and 75th quartiles, and whiskers are the minimum and maximum measured values. * p < 0.05 compared to vehicle control.

STIM2 Is Expressed and Regulates Intracellular [Ca 2+ ] c in NIH 3T3 Cells
STIM2 has been shown to be a regulator of Ca 2+ homeostasis in HeLa, HUVEC, and HEK293T cells [17]. Whether STIM2 plays a similar role in other types of cells remains unclear. The involvement of STIM2 in Ca 2+ homeostasis and signaling was investigated by knocking out its expression in NIH 3T3 fibroblasts using CRISPR-Cas9-mediated genomic editing. STIM2 is expressed in NIH 3T3 cells, and its expression was completely eliminated (KO2-1) by targeting a sequence in exon 2 ( Figure 4A,B). Cells undergoing the same transfection process, but showing no loss of STIM2 expression, were used as controls (WT). The expression of STIM1 was not affected by STIM2 knock-out; however, Orai1 expression was modestly elevated in STIM2 KO2-1 cells ( Figure 4B). Since STIM2 has been reported to be a regulator of basal [Ca 2+ ], we investigated whether STIM2 knock-out would alter cytosolic Ca 2+ homeostasis in unstimulated cells. We found that loss of STIM2 reduced resting [Ca 2+ ] c in STIM2 KO2-1 NIH 3T3 cells (FI340/FI380: 1.27 ± 0.23, n = 294) compared to WT cells (FI340/FI380: 1.40 ± 0.22, n = 172, p < 0.05) ( Figure 4C).

STIM2 Is Expressed and Regulates Intracellular [Ca 2+ ]c in NIH 3T3 Cells
STIM2 has been shown to be a regulator of Ca 2+ homeostasis in HeLa, HUVEC, and HEK293T cells [17]. Whether STIM2 plays a similar role in other types of cells remains unclear. The involvement of STIM2 in Ca 2+ homeostasis and signaling was investigated by knocking out its expression in NIH 3T3 fibroblasts using CRISPR-Cas9-mediated genomic editing. STIM2 is expressed in NIH 3T3 cells, and its expression was completely eliminated (KO2-1) by targeting a sequence in exon 2 ( Figure 4A,B). Cells undergoing the same transfection process, but showing no loss of STIM2 expression, were used as controls (WT). The expression of STIM1 was not affected by STIM2 knock-out; however, Orai1 expression was modestly elevated in STIM2 KO2-1 cells ( Figure  4B). Since STIM2 has been reported to be a regulator of basal [Ca 2+ ], we investigated whether STIM2 knock-out would alter cytosolic Ca 2+ homeostasis in unstimulated cells. We found that loss of STIM2 reduced resting [Ca 2+ ]c in STIM2 KO2-1 NIH 3T3 cells (FI340/FI380: 1.27 ± 0.23, n = 294) compared to WT cells (FI340/FI380: 1.40 ± 0.22, n = 172, p < 0.05) ( Figure 4C). We also measured [Ca 2+ ] ER using D1ER, a fluorescence resonance energy transfer (FRET)-based indicator of [Ca 2+ ] ER [35]. Unlike previous reports, we saw no difference (p > 0.05) between the resting level of [Ca 2+ ] ER in STIM2 knockout (FRET FI535/FI485: 2.8 ± 0.2, n = 101) and control (FRET FI535/FI485: 2.9 ± 0.2, n = 125) cells ( Figure 4D). As an alternative method for measuring the basal level of Ca 2+ in the ER stores, we used Fura-2 to measure the change in the cytosolic [Ca 2+ ] peak following addition of ionomycin, a membrane permeable Ca 2+ ionophore. Cells were bathed in a Ca 2+ -free extracellular solution supplemented with ethylene glycol-bis(β-aminoethyl ether)-N,N,N ,N -tetraacetic acid (EGTA) to prevent the influx of extracellular Ca 2+ across the plasma membrane. The ER is the major Ca 2+ store in the cell; therefore, the normalized maximum Fura-2 FI340/FI380 (∆peak) of the Ca 2+ response following application of ionomycin is an indirect measure of the amount of Ca 2+ sequestered within the lumen of the ER. Using this approach, we found no difference in ER Ca 2+ store content in STIM2 null cells (∆peak: 8.5 ± 2.2, n = 68) compared to WT (∆peak: 8.1 ± 1.6, n = 61) ( Figure 4D). Taken together, our data suggest that knockout of STIM2 modestly reduces basal cytosolic [Ca 2+ ], but not ER Ca 2+ store content in NIH 3T3 cells.

STIM1 Translocation
We used high-resolution fluorescence confocal imaging with STIM1 fused to yellow fluorescent protein (YFP-STIM1) to characterize the distribution of STIM1 before and after store depletion. Confirming previous reports [6,9,44], YFP-STIM1 translocated and formed puncta near the PM following CPA-induced store depletion and returned to a diffuse distribution after the CPA was removed and the ER Ca 2+ stores refilled ( Figure 8A). In contrast, in the NIH 3T3 STIM2 KO2-1 cells, YFP-STIM1 did not form puncta to the same extent as in control cells ( Figure 8B). The puncta in Stim2 null cells were significantly smaller. Quantification of puncta number and area revealed that puncta number per cell was reduced by 46% in the Stim2 null cells, and the average area of each punctum was significantly reduced by 37% (WT: 1.13 ± 0.68 µm 2 , n = 19; KO2-1: 0.71 ± 0.47 µm 2 , n = 18) ( Figure 8D,E). When STIM2 expression was reconstituted by transient overexpression of human STIM2 (hSTIM2), puncta number was fully reconstituted and puncta area was increased by 6% (KO2-1 + hSTIM2: 0.76 ± 0.33 µm 2 , n = 9) ( Figure 8C-E). This suggests that STIM2 is required for efficient translocation and clustering of STIM1 molecules in PAM nanodomains of NIH 3T3 cells.

STIM1 Translocation
We used high-resolution fluorescence confocal imaging with STIM1 fused to yellow fluorescent protein (YFP-STIM1) to characterize the distribution of STIM1 before and after store depletion. Confirming previous reports [6,9,44], YFP-STIM1 translocated and formed puncta near the PM following CPA-induced store depletion and returned to a diffuse distribution after the CPA was removed and the ER Ca 2+ stores refilled ( Figure 8A). In contrast, in the NIH 3T3 STIM2 KO2-1 cells, YFP-STIM1 did not form puncta to the same extent as in control cells ( Figure 8B). The puncta in Stim2 null cells were significantly smaller. Quantification of puncta number and area revealed that puncta number per cell was reduced by 46% in the Stim2 null cells, and the average area of each punctum was significantly reduced by 37% (WT: 1.13 ± 0.68 μm 2 , n = 19; KO2-1: 0.71 ± 0.47 μm 2 , n = 18) ( Figure 8D,E). When STIM2 expression was reconstituted by transient overexpression of human STIM2 (hSTIM2), puncta number was fully reconstituted and puncta area was increased by 6% (KO2-1 + hSTIM2: 0.76 ± 0.33 μm 2 , n = 9) ( Figure 8C-E). This suggests that STIM2 is required for efficient translocation and clustering of STIM1 molecules in PAM nanodomains of NIH 3T3 cells.

STIM2 Facilitates STIM1 Interaction with Orai1
One mechanism by which STIM2 could facilitate STIM1 clustering is through a direct interaction. STIM1-STIM2 interactions have been demonstrated in HeLa cells, human platelets, HEK293 cells, and mouse neurons [17,45,46]. To confirm an interaction of STIM1 with STIM2, we evaluated whether endogenous STIM1 and STIM2 interacts in NIH 3T3 cells using a co-immunoprecipitation assay. Whole cell lysates from control cells with fully filled ER Ca 2+ stores under basal, unstimulated conditions and cells treated with CPA to deplete [Ca 2+ ] ER and activate SOCE were examined. A STIM1-specific antibody was used to pull down STIM1 and immunoprecipitates were analyzed by Western blot for STIM2 and Orai1. STIM2 was detected in the STIM1 immunoprecipitates in both the control and CPA-treated cell lysates; under ER store-depletion conditions, the interaction between STIM1 and STIM2 was enhanced ( Figure 9A). Additionally, the level of Orai1 in the STIM1 immunoprecipitates was increased after depleting ER Ca 2+ stores ( Figure 9A). These findings strongly suggest that STIM1 and Orai1 interact in NIH 3T3 cells similar to that reported in other types of cells [45,[47][48][49]. Since our studies suggested that STIM2 was required for STIM1 puncta formation in the NIH 3T3 cells, we hypothesized that the interaction between STIM1 and Orai1 would be reduced in STIM2 KO cells. To test this hypothesis, we used FRET imaging and found that FRET between YFP-STIM1 and cyan fluorescent protein (CFP)-Orai1 was reduced by 39% in the STIM2 null cells ( Figure 9B,C). Taken together, our data suggests that STIM2 is required for STIM1 puncta formation and efficient clustering at ER-PM junctions where it interacts with and activates Orai1.

STIM2 Facilitates STIM1 Interaction with Orai1
One mechanism by which STIM2 could facilitate STIM1 clustering is through a direct interaction. STIM1-STIM2 interactions have been demonstrated in HeLa cells, human platelets, HEK293 cells, and mouse neurons [17,45,46]. To confirm an interaction of STIM1 with STIM2, we evaluated whether endogenous STIM1 and STIM2 interacts in NIH 3T3 cells using a co-immunoprecipitation assay. Whole cell lysates from control cells with fully filled ER Ca 2+ stores under basal, unstimulated conditions and cells treated with CPA to deplete [Ca 2+ ]ER and activate SOCE were examined. A STIM1-specific antibody was used to pull down STIM1 and immunoprecipitates were analyzed by Western blot for STIM2 and Orai1. STIM2 was detected in the STIM1 immunoprecipitates in both the control and CPA-treated cell lysates; under ER store-depletion conditions, the interaction between STIM1 and STIM2 was enhanced ( Figure 9A). Additionally, the level of Orai1 in the STIM1 immunoprecipitates was increased after depleting ER Ca 2+ stores ( Figure 9A). These findings strongly suggest that STIM1 and Orai1 interact in NIH 3T3 cells similar to that reported in other types of cells [45,[47][48][49]. Since our studies suggested that STIM2 was required for STIM1 puncta formation in the NIH 3T3 cells, we hypothesized that the interaction between STIM1 and Orai1 would be reduced in STIM2 KO cells. To test this hypothesis, we used FRET imaging and found that FRET between YFP-STIM1 and cyan fluorescent protein (CFP)-Orai1 was reduced by 39% in the STIM2 null cells ( Figure 9B,C). Taken together, our data suggests that STIM2 is required for STIM1 puncta formation and efficient clustering at ER-PM junctions where it interacts with and activates Orai1.
STIM2 has been demonstrated to play a role in maintaining basal Ca 2+ homeostasis in mouse neurons, HeLa cells, HUVEC cells and HEK293T cells [17,41]. Our findings confirm this role of STIM2 in NIH 3T3 fibroblasts and αT3 cells, although the levels of cytosolic Ca 2+ reduction in the STIM2 knockout cells were modestly reduced by 10-15% relative to control cells. The effects of silencing Stim2 expression on Ca 2+ homeostasis in other organelles is unclear. Previously, siRNA knockdown of Stim2 revealed that basal [Ca 2+ ] ER in HeLa cells was reduced [17]. In contrast, knockdown of Stim2 in HEK293 cells with siRNA had no effect on thapsigargin-induced release of Ca 2+ [23]. Completely eliminating Stim2 expression in NIH 3T3 and αT3 cells using CRISPR-Cas9 genomic editing has no impact on resting ER Ca 2+ store content. The absence of any effect on ER Ca 2+ levels in resting cells could reflect either differences in the role of Stim2 in the regulation of ER Ca 2+ homeostasis in different types of cells or an effect of partial knockdown of Stim2 compared with a total knockout of the gene's expression using CRISPR-Cas9. Additional work in other types of cells, both in vitro and in vivo are needed to more fully understand how STIM2 regulates Ca 2+ homeostasis in subcellular compartments.
Because STIM2 is a weaker activator of Orai1, we wondered how loss of STIM2 might have such a large impact on SOCE. It was previously shown that at low-stimulus intensities, STIM2 enhances activation of SOCE by promoting STIM1 clustering [46]. Additionally, in rat hippocampal neurons, STIM1 did not co-localize with Orai1 in the absence of STIM2 [55]. Given these data, we hypothesized that in Stim2 null cells, SOCE was reduced due to an inability of STIM1 to interact fully with Orai1. Indeed, we show YFP-STIM1 does not efficiently translocate and form puncta when overexpressed in STIM2 KO cells as quantified by reduced puncta number and area, suggesting a role for STIM2 in STIM1 puncta formation. Interestingly, we found this to be true even following maximal store-depletion induced by SERCA inhibition with CPA, not just at low stimulus intensities. Furthermore, our FRET imaging studies in living cells showed that STIM1 interaction with Orai1 was impaired in Stim2 null cells. Taken together, these data suggest a critical role for STIM2 in facilitating the association of functional STIM1-Orai1 complexes.
Differences between our work and previously published data regarding a threshold of activation for STIM2 recruitment of STIM1 to ER-PM junctions is likely explained by differences in experimental approach used to silence Stim2 expression. Many of the early studies used siRNA knockdown of STIM2 [2,21,23,37,42], yielding varying levels of efficacy, which could significantly impact the interpretation of the data. It is likely that partial knockdown of STIM2 expression may not be sufficient to reveal a role of STIM2 in SOCE. Our work, using CRISPR-Cas9 technology to completely ablate Stim2 expression, indicated that STIM2 is required for STIM1 translocation, the interaction of STIM1 with Orai1, and activation of SOCE. We propose a model in which STIM2 is required for STIM1 puncta formation and activation of Orai1 ( Figure 10). We suggest that when the ER Ca 2+ stores are full, STIM1 and STIM2 are located diffusely within the ER. Following stimulation and store-depletion, STIM1 oligomerizes and associates with STIM2 through the conserved SAM and CC domains. The STIM2 EF-SAM domain is more stable than that of STIM1 [19], which may be important for its ability to stabilize large STIM1 oligomers. Furthermore, the lysine-rich (K-rich) domain present in the C-terminus of both STIM proteins binds to PM phosphatidylinositol 4,5-bisphosphate (PIP 2 ) to stabilize STIM at ER-PM junctions [6,56,57]. STIM2 was shown to have a higher affinity for PM lipids than STIM1, suggesting that it may be a stronger stabilizer of PAM functional integrity and critical for supporting STIM1-dependent Ca 2+ signaling in these junctions [57]. This may help STIM2 serve as an anchoring protein. Through strong electrostatic interactions with the PM, STIM2 is able to anchor and stabilize large STIM1 clusters in close proximity to Orai1. Additional studies will be necessary to better understand why in the absence of STIM2, STIM1 is not able to form large clusters and is inefficient at gating Orai1, resulting in the reduced SOCE. ER Ca 2+ stores are full, STIM1 and STIM2 are located diffusely within the ER. Following stimulation and store-depletion, STIM1 oligomerizes and associates with STIM2 through the conserved SAM and CC domains. The STIM2 EF-SAM domain is more stable than that of STIM1 [19], which may be important for its ability to stabilize large STIM1 oligomers. Furthermore, the lysine-rich (K-rich) domain present in the C-terminus of both STIM proteins binds to PM phosphatidylinositol 4,5-bisphosphate (PIP2) to stabilize STIM at ER-PM junctions [6,56,57]. STIM2 was shown to have a higher affinity for PM lipids than STIM1, suggesting that it may be a stronger stabilizer of PAM functional integrity and critical for supporting STIM1-dependent Ca 2+ signaling in these junctions [57]. This may help STIM2 serve as an anchoring protein. Through strong electrostatic interactions with the PM, STIM2 is able to anchor and stabilize large STIM1 clusters in close proximity to Orai1. Additional studies will be necessary to better understand why in the absence of STIM2, STIM1 is not able to form large clusters and is inefficient at gating Orai1, resulting in the reduced SOCE. Figure 10. Model illustrating the importance of STIM2 in the clustering and recruitment of STIM1 to ER-PM junctions where it interacts with Orai1. At rest when the intracellular Ca 2+ stores are full, the EF-hand domain of STIM1 and STIM2 bind Ca 2+ and maintain an inactive confirmation (left). In WT cells containing STIM1 and STIM2, Ca 2+ store-depletion causes Ca 2+ dissociation from the EF-hand, initiating a destabilization-coupled oligomerization event. STIM2 provides a building block for large multimers of STIM1 to form. Given the STIM2 lysine-rich domain has a higher affinity for PM phospholipids, it likely plays a role in recruiting and stabilizing STIM1 at ER-PM junctions, where it interacts with and activates Orai1 (center). In the absence of STIM2, STIM1 cannot form large oligomers and is inefficient at interacting with and gating Orai1 (right).
In conclusion, our data demonstrate that STIM2 is a critical regulator of SOCE in NIH 3T3 fibroblasts and αT3 cells. Given the importance of STIM2 in facilitating SOCE, it may be an attractive target for SOCE inhibition in diseases associated with enhanced SOCE. For example, several cancers have been associated with upregulated expression of STIM1, STIM2, or Orai1 and increased SOCE [58,59]. STIM1 has a tumor growth promoting role in patients with breast cancer and cervical cancer [54,60]. In a genome-wide gene expression analysis of 20 primary glioblastoma samples, Stim2 expression was upregulated [61]. Similarly, knockdown of STIM1 or Orai1 in rat and human glioblastoma cells inhibited tumor cell proliferation and promoted apoptosis [62]. Caution must be taken, however, when targeting STIM2 for therapeutic intervention in disease. Downregulation of Stim2 expression is not always associated with decreased SOCE. In HT29 colon cancer cells, SOCE was significantly enhanced despite reduced Stim2 expression [32]. This is likely a result of changes increased expression of TRPC1, Orai1, and STIM1. In comparison, reduction of Stim2 in normal mucosal cells reduced SOCE and Ca 2+ store content and promoted apoptosis resistance, suggesting a role for downregulated Stim2 expression in tumor cell survival. Therefore, therapeutic interventions using Stim2 as a molecular target are likely going to be cell type-specific and attention to effects of modifying Stim2 expression on other proteins that regulate SOCE will be required.

Orai1 Closed
Orai1 Open PM

Resting
Store-depletion with STIM2 Store-depletion no STIM2 Ca 2+ Cytosol STIM1 STIM2 Figure 10. Model illustrating the importance of STIM2 in the clustering and recruitment of STIM1 to ER-PM junctions where it interacts with Orai1. At rest when the intracellular Ca 2+ stores are full, the EF-hand domain of STIM1 and STIM2 bind Ca 2+ and maintain an inactive confirmation (left). In WT cells containing STIM1 and STIM2, Ca 2+ store-depletion causes Ca 2+ dissociation from the EF-hand, initiating a destabilization-coupled oligomerization event. STIM2 provides a building block for large multimers of STIM1 to form. Given the STIM2 lysine-rich domain has a higher affinity for PM phospholipids, it likely plays a role in recruiting and stabilizing STIM1 at ER-PM junctions, where it interacts with and activates Orai1 (center). In the absence of STIM2, STIM1 cannot form large oligomers and is inefficient at interacting with and gating Orai1 (right).
In conclusion, our data demonstrate that STIM2 is a critical regulator of SOCE in NIH 3T3 fibroblasts and αT3 cells. Given the importance of STIM2 in facilitating SOCE, it may be an attractive target for SOCE inhibition in diseases associated with enhanced SOCE. For example, several cancers have been associated with upregulated expression of STIM1, STIM2, or Orai1 and increased SOCE [58,59]. STIM1 has a tumor growth promoting role in patients with breast cancer and cervical cancer [54,60]. In a genome-wide gene expression analysis of 20 primary glioblastoma samples, Stim2 expression was upregulated [61]. Similarly, knockdown of STIM1 or Orai1 in rat and human glioblastoma cells inhibited tumor cell proliferation and promoted apoptosis [62]. Caution must be taken, however, when targeting STIM2 for therapeutic intervention in disease. Downregulation of Stim2 expression is not always associated with decreased SOCE. In HT29 colon cancer cells, SOCE was significantly enhanced despite reduced Stim2 expression [32]. This is likely a result of changes increased expression of TRPC1, Orai1, and STIM1. In comparison, reduction of Stim2 in normal mucosal cells reduced SOCE and Ca 2+ store content and promoted apoptosis resistance, suggesting a role for downregulated Stim2 expression in tumor cell survival. Therefore, therapeutic interventions using Stim2 as a molecular target are likely going to be cell type-specific and attention to effects of modifying Stim2 expression on other proteins that regulate SOCE will be required.
1% SDS (pH 8.3)). The samples were then heated at 37 • C for 30 min, centrifuged for 60 s at 13,000 rpm, and the supernatant was used for Western blotting as described above.

Generation and Analysis of STIM2 Knock-Out (KO) Cell Lines
The CRISPR-Cas9 system was used to target exons in the mouse Stim2 gene. Oligonucleotides targeting exon 2 (5 -CAAGGACGGCGGGATCGAAG-3 ) were annealed and ligated into AflII-linearized gRNA vector (Addgene, Cambridge, MA, USA). Alternatively, exon 8 (5 -GATGCAGCTAGCCATCGCTA-3 ) was targeted to confirm our findings from our exon 2 KO. CRISPR target sequences were searched using NCBI BLAST to verify specificity. Following molecular cloning to insert our target sequences, the CRISPR gRNA vectors were sequenced for verification. Lipofectamine 2000 was used to transfect cells with a mixture of the Stim2-targeted gRNA construct, and vectors encoding hCas9 (Addgene, Cambridge, MA, USA) and EGFP-c1. In the αT3 cells, the same exon 2 target sequence was used, but cloned into a polycistronic vector (pCas-Guide-EF1a-GFP) that also contained hCas9 and GFP (OriGene, Rockville, MD, USA, Cat. No. GE100018). In the polycistronic vector we also cloned in a 20 base-pair scramble sequence to generate controls (5 -GTCGCTTGGGCGAGAGTAAG-3 ). Two days post-transfection, EGFP-expressing cells were selected by fluorescence-activated cell sorting and plated one cell per well in a 96-well plate. Colonies were expanded and screened for knock-out of STIM2 expression by Western immunoblotting with anti-STIM2 (Cell Signaling Technology, Danvers, MA, USA, Cat. No. 4917) as described above.

Flex Station
The SOCE inhibitor studies and NMG experiments were performed in a 96-well format in a FlexStation 3 multi-mode microplate reader (Molecular Devices, LLC., San Jose, CA, USA). NIH 3T3 cells (40,000 per well) and αT3 cells (50,000 per well) were seeded in a black-walled 96-well plate. Twenty-four hours later, cells were loaded with Fura-2 for 30 min at room temperature (SES, 1 µM Fura-2, 0.01% Pluronic ® F-127). The loading solution was removed and replaced with SES. [Ca 2+ ] c measurements were made using a FlexStation 3 plate reader controlled with SoftMax Pro software (Molecular Devices). Cells were excited at 335 nm and 375 nm, and the emitted light was detected at 505 nm using a 435-nm dichroic mirror. Test solutions were added to individual wells using the automated injection function of the FlexStation 3 at time points indicated in the figures. Data are expressed as the ratio of FI335/FI375 or relative FI335/FI375 (fold-change in FI335/FI375 relative to basal, resting FI335/FI375).

ER Ca 2+ Imaging
Cells were transfected with a plasmid construct encoding D1ER, a fluorescence resonance energy transfer (FRET)-based indicator of [Ca 2+ ] ER , by electroporation (Neon Transfection System, Thermo Fisher Scientific, Waltham, MA, USA). A suspension of NIH 3T3 cells (5 × 10 6 cells/mL) was electroporated at 1350 V with two 20 ms pulses. Parameters for the αT3 cell electroporation were 1500 volts, 1 pulse, 20 ms. Cells were seeded and cultured on glass coverslips for 48 h at 37 • C, placed into a heated microperifusion chamber mounted on the stage of an inverted microscope (Olympus IX81) equipped with a charge-coupled device camera (ImagEM X2-CCD, Hamamatsu Photonics, Bridgewater, NJ, USA). Cells were superfused with SES or KRH2 at 37 • C and visualized with a 40× oil immersion objective. D1ER was excited at 440 nm and emission intensities were measured at 485 nm (FRET donor) and 535 nm (FRET acceptor). MetaFluor software (Molecular Devices LLC., San Jose, CA, USA) was used for image acquisition and analysis. Data are expressed as the fold-change in ratio of FRET acceptor and donor emission intensities (FRET FI535/FI485) normalized to resting, unstimulated values.
As an alternative strategy to indirectly measure ER Ca 2+ store content, NIH 3T3 cells on coverslips were loaded with Fura-2 and images were acquired as described above. A SES/0Ca 2+ buffer with 10 µM EGTA was perifused onto the coverslip before the addition of ionomycin to eliminate extracellular Ca 2+ influx. Ionomycin was added to a final concentration of 2 µM and the subsequent change in [Ca 2+ ] c was measured.

FRET Measurements of Protein-Protein Interactions
To measure the interaction of STIM1 and Orai1, cells were co-transfected with YFP-STIM1 (FRET acceptor) and CFP-Orai1 (FRET donor) using electroporation as described above. Quantitative FRET imaging was performed as described above. The WT cells served as our biological control, showing differences in the magnitude of FRET between Orai1-CFP and STIM1-YFP in WT cells compared to KO2-1. Additionally, a vehicle control (DMSO) was used to show no change in FRET when perifused onto cells instead of CPA. Data are expressed as the fold-change in the background subtracted ratio of FRET acceptor and donor emission intensities (FRET FI535/FI485 nm) normalized to resting values. Resting FI535/FI485 values in KO2-1 and WT were not different.

Morphometric Analysis of STIM1 Localization
Cells were transfected with YFP-STIM1 in WT and KO2-1 cells. To reconstitute STIM2 expression in KO2-1 cells, YFP-STIM1 was co-expressed with CFP-STIM2. Two days post-transfection, a time-lapse series of images were acquired in which cells were perifused with test solutions at 37 • C. Cells were initially bathed in normal, 1 mM Ca 2+ extracellular solution to establish a baseline distribution, and then superfused with a Ca 2+ -free solution supplemented with 20 µM cyclopiazonic acid (CPA) to discharge and prevent refilling of intracellular Ca 2+ stores. To restore ER Ca 2+ levels, cells were bathed in the normal extracellular solution without CPA. Confocal images acquired immediately before the reconstitution of extracellular Ca 2+ and CPA washout were background subtracted and processed using 2D deconvolution. Morphometric analysis was used to count the number and determine the area of individual puncta in each cell (MetaMorph ® Microscopy Automation & Image Analysis Software, Version 7.0, Molecular Devices, LLC., San Jose, CA, USA).

Reagents
Fura-2 acetoxymethylester and ionomycin were purchased from Thermo Fisher Scientific and CalBiochem, respectively.
ML-9 and SKF96365 hydrochloride were obtained from Sigma-Aldrich Co. Cyclopiazonic acid (CPA) was obtained from Millapore Sigma. Orai1-E106A, CFP-Orai1 and YFP-STIM2 were obtained from Addgene.org. YFP-STIM1 was a generous gift from Tobias Meyer (Stanford University). D1ER was provided by the late Roger Y. Tsien (University of California San Diego).

Statistical Analysis
Our data was analyzed and revealed nearly normal to normal distribution, equal variance, and were independent measurements; therefore, we used unpaired Student's t-test assuming equal variance for intergroup comparisons (p < 0.05 was considered statistically significant), as detailed in the figure legends. Data traces are plotted as the mean ± SEM. Box and whisker plots show mean (open square), median (solid line), 25th and 75th percentiles (box), and minimum and maximum measurements (whiskers).