Cross-talk between N-terminal and C-terminal domains in stromal interaction molecule 2 (STIM2) determines enhanced STIM2 sensitivity

Store-operated Ca2+ entry (SOCE) is a ubiquitous pathway for Ca2+ influx across the plasma membrane (PM). SOCE is mediated by the endoplasmic reticulum (ER)-associated Ca2+-sensing proteins stromal interaction molecule 1 (STIM1) and STIM2, which transition into an active conformation in response to ER Ca2+ store depletion, thereby interacting with and gating PM-associated ORAI1 channels. Although structurally homologous, STIM1 and STIM2 generate distinct Ca2+ signatures in response to varying strengths of agonist stimulation. The physiological functions of these Ca2+ signatures, particularly under native conditions, remain unclear. To investigate the structural properties distinguishing STIM1 and STIM2 activation of ORAI1 channels under native conditions, here we used CRISPR/Cas9 to generate STIM1−/−, STIM2−/−, and STIM1/2−/− knockouts in HEK293 and colorectal HCT116 cells. We show that depending on cell type, STIM2 can significantly sustain SOCE in response to maximal store depletion. Utilizing the SOCE modifier 2-aminoethoxydiphenyl borate (2-APB), we demonstrate that 2-APB–activated store-independent Ca2+ entry is mediated exclusively by endogenous STIM2. Using variants that either stabilize or disrupt intramolecular interactions of STIM C termini, we show that the increased flexibility of the STIM2 C terminus contributes to its selective store-independent activation by 2-APB. However, STIM1 variants with enhanced flexibility in the C terminus failed to support its store-independent activation. STIM1/STIM2 chimeric constructs indicated that coordination between N-terminal sensitivity and C-terminal flexibility is required for specific store-independent STIM2 activation. Our results clarify the structural determinants underlying activation of specific STIM isoforms, insights that are potentially useful for isoform-selective drug targeting.

In nonexcitable cells, cytosolic Ca 2ϩ concentrations are controlled through release from intracellular Ca 2ϩ stores of the endoplasmic reticulum (ER) 3 and Ca 2ϩ influx across the plasma membrane (PM) through the ubiquitous store-operated Ca 2ϩ entry (SOCE) pathway (1). SOCE is required for the acute refilling of ER Ca 2ϩ stores and sustaining downstream signaling to gene transcription in most cell types (2)(3)(4)(5). SOCE is mediated by the ER Ca 2ϩ sensors stromal interaction molecules (STIM1 and STIM2), which directly bind, cross-link, and gate plasma membrane ORAI1 channels in response to store depletion. Extensive investigation into the molecular interactions between STIM1 and ORAI1 has revealed STIM1 as a potent activator of ORAI1 channels through a strategically positioned phenylalanine (Phe-394) located within the STIM-ORAIactivating region (SOAR1) in the STIM1 C terminus (6 -8). Furthermore, the N-terminal EF-hand domain of STIM1 located within the ER lumen has a high affinity for Ca 2ϩ , requiring significant depletion of ER Ca 2ϩ stores for the transition of STIM1 into an active conformation (9,10). In contrast, the structurally homologous STIM2 is a relatively weak activator of ORAI1, in which the residue equivalent to Phe-394 of STIM1 is a leucine (Leu-485) within the SOAR domain of STIM2 (SOAR2) that renders it a partial agonist (6). The luminal N-terminal EF-hand domain of STIM2 has a lower affinity for Ca 2ϩ compared with STIM1, allowing STIM2 to sense minor reductions in ER Ca 2ϩ concentrations (11). As a result, a significant pool of STIM2 protein appears to be constitutively active and bound to ORAI1 channels, possibly as a mechanism to maintain basal cytosolic and ER Ca 2ϩ concentrations (11,12). Interestingly, a previous study proposed that the short N-terminal peptide before the EF-hand domain in STIM2 functions as "brake" to ensure negative regulation of STIM2 constitutive activity and prevent significant enhancement of basal cytosolic Ca 2ϩ concentration (13). A recent report further showed that the E470G mutation within the SOAR2 domain of STIM2 (Glu-470 is equivalent to Gly-379 in STIM1) renders STIM2 a fast and robust activator of ORAI1, whereas the opposite mutation in STIM1 only weakly activates ORAI1 (14). The same group showed that the transmembrane (TM) domain of STIM2 contributes to its slow activation kinetics compared with STIM1 (14). Although STIM1 has been shown to drive the majority of SOCE in a variety of cell types in response to agonist-induced store depletion (9,(15)(16)(17), the physiological function of STIM2 in this process is not clear. Further adding to the complexity of this system is recent evidence demonstrating that STIM2 plays a role in inducing the active conformation of STIM1 and regulating its coupling with ORAI1 channels under conditions of replete ER Ca 2ϩ stores (18,19). The physiological implications of this mechanism, particularly how endogenous STIM proteins utilize this molecular coordination to drive cell function, are unknown. Additionally, the structural differences between STIM1 and STIM2 that determine their differential response to varying strengths of agonist and the distinct roles of STIM1 and STIM2 in cell physiology, especially under native conditions, are largely unknown.
Multiple classes of pharmacological compounds that either enhance or inhibit SOCE activity have been identified (20,21). Among these SOCE channel modifiers, the compound 2-aminoethoxydiphenyl borate (2-APB) has been well characterized and widely utilized (22)(23)(24). 2-APB displays a unique biphasic effect on ORAI1 activity, potentiating it at low concentrations (1-10 M) and inhibiting it at high concentrations (ϳ30 -50 M) (25). High concentrations of 2-APB have been demonstrated to: 1) inhibit several constitutively active STIM-independent ORAI1 channel mutants, whereas potentiating WT ORAI3, suggesting that 2-APB acts directly on ORAI channel pore; 2) strengthen intramolecular interaction within the STIM1 C terminus, between the coiled-coil 1 (CC1) domain and SOAR1, thus preventing STIM1 unfolding in response to store depletion. This would result in a reduction in STIM1 oligomerization and puncta formation that would reduce functional interactions with ORAI1 (22, 26 -31). However, the ability of 2-APB at low concentrations to selectively increase Ca 2ϩ influx through ORAI1 channels remains enigmatic.
Although 2-APB is relatively nonspecific and can affect a diverse array of ion channels, a clear understanding of how 2-APB interacts with and regulates ORAI and STIM proteins under native levels of expression would be crucial for the future development of potent and selective SOCE modifying compounds (32)(33)(34). Here, we generated multiple cell lines using CRISPR/Cas9 technology that are devoid of individual STIM proteins (STIM1 Ϫ/Ϫ , STIM2 Ϫ/Ϫ ) as well as double STIM1 and STIM2 knockout cells (STIM1/2 Ϫ/Ϫ ). Our results demonstrate that store-independent activation of SOCE with low concentrations of 2-APB (10 M) occurs exclusively through endogenous STIM2. Through the use of chimeric STIM proteins and STIM mutant constructs, we clearly demonstrate that the effects of 2-APB on STIM2 requires the combination of increased flexibility of the STIM2 C terminus and increased sensitivity of its N-terminal EF-hand. Thus, the use of low concentrations of 2-APB as a pharmacological tool to selectively drive native STIM2-mediated Ca 2ϩ entry provides a novel strategy to dissect the physiological functions of STIM2 under homeostatic and endogenous conditions. Furthermore, development of 2-APB-derived analogues with increased specificity and potency for STIM2 will selectively target cellular pathways that rely on STIM2 in both healthy and disease conditions.

Results
Generation and characterization of STIM1 ؊/؊ , STIM2 ؊/؊ , and STIM1/2 ؊/؊ knockout cell lines We utilized CRISPR/Cas9 technology to delete each STIM protein individually and in combination in HEK293 cells. Western blot analysis of HEK293 STIM1 Ϫ/Ϫ and STIM2 Ϫ/Ϫ cell lines demonstrated no compensatory up-regulation of STIM1 or STIM2 proteins in response to individual STIM knockout, whereas STIM1/2 Ϫ/Ϫ cells were devoid of both proteins (Fig. 1,  A and B). Fura-2 Ca 2ϩ imaging was performed in each HEK293 knockout cell line to measure SOCE in response to store depletion with the sarco/endoplasmic reticulum Ca 2ϩ -ATPase (SERCA) pump inhibitor thapsigargin ( Fig. 1, C and D). HEK293 STIM1 Ϫ/Ϫ cells exhibit substantial inhibition in peak SOCE (to ϳ10% of WT), whereas STIM2 Ϫ/Ϫ cells showed only minor reduction in peak SOCE (ϳ84% of WT). Knockout of both STIM1 and STIM2 (HEK293 STIM1/2 Ϫ/Ϫ cells) results in essentially complete abrogation of SOCE (ϳ3% of WT) upon re-addition 2 mM Ca 2ϩ to the extracellular medium. To investigate cell type-dependent contributions of STIM1 versus STIM2 to Ca 2ϩ signals, we also generated individual STIM1 and STIM2 knockout of the colorectal cancer cell line HCT116 and documented knockouts with Western blots (Fig. 1, E and  F). Ca 2ϩ imaging experiments demonstrated a significant reduction in peak SOCE in HCT116 STIM1 Ϫ/Ϫ cells (ϳ24% of WT) (Fig. 1, G and H). Interestingly, HCT116 STIM2 Ϫ/Ϫ cells also showed substantial reduction in SOCE (ϳ44% of WT) despite having STIM1 protein present. These results suggest that STIM2 plays differential, cell type-specific role in supporting SOCE in response to maximal ER store depletion.

2-APB activates store-independent Ca 2؉ entry exclusively through STIM2
Using our newly generated HEK293 and HCT116 STIM knockout cell lines, we investigated the effects of low (10 M) and high (50 M) 2-APB under conditions where internal Ca 2ϩ stores were replete. To address potential off target effects of CRISPR/Cas9, we generated additional STIM1 and STIM2 knockout clones in both cell lines using multiple guide RNA sequences (Fig. 2, A and F). Our results presented in detail below show that STIM2, but not STIM1, is crucial for 2-APBactivated Ca 2ϩ entry. To strengthen this argument, we present data obtained from 4 STIM2 Ϫ/Ϫ HCT116 cell clones representing 2 independent clones each obtained from two independent guide RNA (e.g. STIM2 Ϫ/Ϫ g1.1 corresponds to clone 1 from guide RNA 1 etc.; Fig. 2F). Stimulation of WT HEK293 and HEK STIM1 Ϫ/Ϫ cells with 10 M 2-APB in the presence of 2 mM extracellular Ca 2ϩ demonstrated a gradual and sustained increase in cytosolic Ca 2ϩ concentrations (Fig. 2, B and C). Importantly, this potentiation was not observed in HEK STIM2 Ϫ/Ϫ or STIM1/2 Ϫ/Ϫ cell lines, strongly arguing that Structural determinants of enhanced STIM2 sensitivity STIM2 is the target of 2-APB store-independent activation of Ca 2ϩ entry. Similar to HEK293 cells, stimulation of HCT116 WT and STIM1 Ϫ/Ϫ cells with 10 M 2-APB demonstrated a small yet sustained Ca 2ϩ entry, which was not observed in HCT116 STIM2 Ϫ/Ϫ cells (Fig. 2, G and H). Addition of 50 M 2-APB to different variants of HEK293 cells showed a marginal and transient increase in cytosolic Ca 2ϩ (Fig. 2, D and E). Surprisingly, stimulation of WT HCT116 cells with 50 M 2-APB showed a more pronounced, rapid and transient increase in cytosolic Ca 2ϩ , and this effect was further enhanced in HCT116 STIM1 Ϫ/Ϫ cells (Fig. 2, I and J). This rapid potentiation with 50 M 2-APB was not observed in HCT116 STIM2 Ϫ/Ϫ cells (Fig. 2, I and J). These results suggest that 2-APB activates endogenous STIM2, but not STIM1, in both cell lines in a store-independent manner and that the more potent activation of Ca 2ϩ entry by 50 M 2-APB in HCT116 cells results from the more prominent role STIM2 plays in SOCE of HCT116 cells.
Depletion of intracellular Ca 2ϩ stores induces rapid puncta formation of STIM1 and induces clustering with ORAI1 channels (7,35). High concentrations of 2-APB (50 M) inhibit the formation of STIM1 puncta by clamping the STIM1 C terminus

Low concentrations of 2-APB potentiate SOCE exclusively through STIM2
We next examined the effects of 2-APB on the potentiation and inhibition of SOCE activated through agonist stimulation. For reasons that will become clear below, we used 100 M carbachol with 2 mM extracellular Ca 2ϩ or 150 M ATP with 10 mM extracellular Ca 2ϩ for HEK293 cells and 300 M ATP with 10 mM extracellular Ca 2ϩ for HCT116 cells. Stimulation with carbachol followed by re-addition of Ca 2ϩ to the extracellular

Structural determinants of enhanced STIM2 sensitivity
medium induced SOCE in WT HEK293 cells. Although HEK293 STIM2 Ϫ/Ϫ cells showed only partial inhibition of SOCE, minimal SOCE was detected in STIM1 Ϫ/Ϫ , and essentially no SOCE was observed in STIM1/2 Ϫ/Ϫ cells (Fig. 5A). The same protocol using stimulation with ATP showed marginal Ca 2ϩ entry in all three knockout cell lines, STIM1 Ϫ/Ϫ , STIM2 Ϫ/Ϫ , and STIM1/2 Ϫ/Ϫ (Fig. 5B), suggesting that ATP is a weak agonist that relies on both STIM1 and STIM2 for SOCE activation. Indeed, 10 mM external Ca 2ϩ was used to enhance the driving force and detect a Ca 2ϩ signal in WT HEK293 cells. In all conditions, however, subsequent addition of 10 M 2-APB consistently induced rapid potentiation of SOCE in STIM1 Ϫ/Ϫ cells, whereas no potentiation was observed in STIM2 Ϫ/Ϫ or STIM1/2 Ϫ/Ϫ cells (Fig. 5, A and B). SOCE was inhibited by the sequential addition of 50 M 2-APB in all cell lines (Fig. 5, A and  B). WT HEK293 cells showed marginal potentiation with 10 M 2-APB when SOCE was stimulated with carbachol ( Fig. 5A), likely because carbachol causes maximal depletion of ER stores and maximal activation of SOCE compared with ATP (Fig. 5B). Similar results were observed with ATP in HCT116 cells. Supramaximal ATP concentrations (300 M with 10 mM Ca 2ϩ extracellular; 150 M ATP did not yield consistent SOCE activation) activated moderate SOCE in WT HCT116 cells, with marginal SOCE detected in STIM1 Ϫ/Ϫ and STIM2 Ϫ/Ϫ cells (Fig. 5C). Addition of 10 M 2-APB potentiated SOCE in WT HCT116 and STIM1 Ϫ/Ϫ cells, but not in STIM2 Ϫ/Ϫ cells. This potentiation was rapidly inhibited following addition of 50 M 2-APB (Fig. 5C). Despite repeated attempts with various protocols, HCT116 cell lines did not respond to carbachol stimulation.

Flexibility of the STIM2 C terminus is critical to 2-APB selectivity
Under conditions of high ER Ca 2ϩ concentrations, the C terminus of STIM1 and STIM2 are maintained in an inactive and clamped confirmation characterized by strong molecular interactions between the CC1 and CC3 domains, which shield SOAR domains from ORAI1 (14, 18, 36 -39). Depletion of ER stores and loss of Ca 2ϩ binding to the luminal EF-hand domains of STIM1 and STIM2 triggers rearrangement of the N and C terminus, opening the CC1/CC3 clamp, and exposing either SOAR1 or SOAR2 for gating and activation of ORAI1 channels (2, 4, 6, 37, 38, 40 -42). Accordingly, 2-APB-induced store-independent Ca 2ϩ entry is mediated through proper interaction and gating of SOAR2 with ORAI1 channels as the dominantnegative splice variant STIM2.1, containing an 8 amino acid insertion within SOAR2 (43,44), demonstrated no increase in cytosolic Ca 2ϩ concentrations in response to 10 M 2-APB (Fig.  6, A and B). In conjunction with the SOAR domains (Fig. 6C), studies have described key residues within the C-terminal CC1

Structural determinants of enhanced STIM2 sensitivity
and CC3 domains that contribute to the induction of the active conformation of STIM1 and STIM2 (37,40). Mutation of residues located within the ORAI1 activating small fragment (OASF1 (37,40), Fig. 6C) region of STIM1 have been identified to either stabilize (R426L) or disrupt (L251S) CC1/CC3 interactions (Fig. 6C). Comparison of corresponding mutations in STIM2 that stabilize (R517L) or disrupt (L342S/L507S/L514S, CCmt) CC1/CC3 interactions suggest that the OASF region of STIM2 (OASF2) adopts a weaker CC1/CC3 interaction generating a more flexible conformation of OASF2 compared with OASF1 (19). Utilizing intramolecular FRET-based OASF1 and OASF2 sensors expressed in HEK293 STIM1/2 Ϫ/Ϫ cells, we confirm previous findings of significantly reduced intramolecular FRET signal of the OASF2 region of STIM2 compared with OASF1 of STIM1 (Fig. 6D). We further show that OASF1 and OASF2 FRET sensors with mutations predicted to stabilize or disrupt CC1/CC3 interactions behave as expected. The OASF1-L251S mutant predicted to disrupt STIM1 C-terminal interactions displayed significantly reduced intramolecular FRET signal compared with OASF2 (Fig. 6E). Conversely, the stabilizing mutant OASF1-R426L displayed a FRET signal comparable with OASF1. Importantly, the triple L342S/L507S/ L514S (CCmt) OASF2 mutant displayed significantly decreased FRET signal compared with OASF2 (Fig. 6E), suggesting even further flexibility of this mutant. The stabilized OASF2-R517L mutant demonstrated significantly higher FRET signal than OASF2, which was still substantially less than OASF1 (Fig. 6E). We next tested the ability of 10 M 2-APB to induce changes in intramolecular FRET of OASF1 and OASF2 constructs and observed marginal FRET changes with all different OASF variants (Fig. 6, F and G; see H and I for zoom-in normalized FRET), suggesting that any rearrangements of C-terminal STIM caused by 2-APB are likely very subtle. These results are consistent with data from the Romanin group (37) showing that, without overexpression of ORAI1, even high concentrations (75 M) of 2-APB caused a marginal decrease in OASF1 intramolecular FRET.
We next evaluated the effects of CC1/CC3 stabilization or disruption mutations on 2-APB-mediated Ca 2ϩ entry by expressing these mutations within the context of full-length STIM1 and STIM2 proteins in HEK293 STIM1/2 Ϫ/Ϫ cells. Similar to overexpression of STIM2, both the flexible CCmt-STIM2 and stabilized R517L-STIM2 mutants displayed extensive preformed puncta (Fig. 7A). Although all STIM2 mutants showed reorganization of baseline puncta in response to the change in shape of cells, 2-APB did not lead to significant enhancement of puncta size (Fig. 7B). Stimulation of STIM2expressing cells with 10 M 2-APB induced rapid and sustained store-independent Ca 2ϩ entry (Fig. 7, C and D), as shown in Fig.  3, C and D. Interestingly, the flexible CCmt-STIM2 displayed

Structural determinants of enhanced STIM2 sensitivity
substantially increased Ca 2ϩ entry compared with STIM2, whereas the stabilized R517L-STIM2 supported Ca 2ϩ entry that was slightly reduced compared with STIM2 (Fig. 7, C and  D). Similar results were obtained when STIM2 mutants were stimulated with 50 M 2-APB (Fig. 7, E and F). However, the stabilized R517L-STIM2 mutant displayed nearly identical Ca 2ϩ entry compared with STIM2 (Fig. 7, E and F).
STIM1 and the stabilized R426L-STIM1 mutant expressed in STIM1/2 Ϫ/Ϫ cells displayed tubular localization throughout the cell, whereas the flexible L251S-STIM1 had extensively preformed puncta (Fig. 8A). Stimulation of STIM1 with 10 M 2-APB induced minimal store-independent Ca 2ϩ entry, which was completely abolished in the stabilized R426L-STIM1 mutant (Fig. 8, C and D). The flexible L251S-STIM1 mutant showed preformed puncta and constitutive Ca 2ϩ activity (Fig.  8B) as previously reported (37). Unexpectedly, however, it supported minimal Ca 2ϩ entry in response to 10 M 2-APB that was comparable with STIM1 (Fig. 8, C and D). In response to 50 M 2-APB, STIM1 and the flexible L251S-STIM1 mutants showed similar patterns of small initial Ca 2ϩ entry to those seen on stimulation with 10 M 2-APB, whereas the stabilized R426L-STIM1 showed essentially no response (Fig. 8, E and F). Furthermore, basal activity of the L251S-STIM1 mutant was subsequently inhibited by 50 M 2-APB (Fig. 8E). Our results thus far suggest that whereas the increased flexibility of the STIM2 C terminus plays a key role in driving Ca 2ϩ entry in response to 2-APB and weak agonists, enhanced C-terminal flexibility alone cannot rescue the activation of STIM1 by 2-APB.

Structural determinants of enhanced STIM2 sensitivity
Recent reports have demonstrated that STIM2 remodels the STIM1 C terminus under conditions of high ER Ca 2ϩ concentrations to allow interactions with ORAI1 under low stimulus intensities (18,19). We examined this effect by co-expression of STIM1 and STIM2 in STIM1/2 Ϫ/Ϫ cells. Interestingly, the expression of both STIM1 and STIM2 led to significant constitutive activity by comparison to either STIM expressed alone (Fig. 8G). Stimulation with either 10 M (Fig. 8, H and I) or 50 M (Fig. 8, J and K) 2-APB activated significant Ca 2ϩ entry in cells expressing STIM2 or co-expressing both STIM proteins but not in cell expressing STIM1 alone.

Coordination between STIM2 N and C termini is required for 2-APB sensitivity
We next determined the contribution of the STIM2 N terminus in mediating store-independent Ca 2ϩ entry by constructing two chimeric STIM constructs: a chimera of STIM1 N terminus with the STIM2 TM and C terminus (S1N-S2C) and another of STIM2 N terminus with the STIM1 TM and C terminus (S2N-S1C; Fig. 9A). Interestingly, expression of the S1N-S2C chimera in HEK293 STIM1/2 Ϫ/Ϫ cells showed preformed puncta throughout the cell (Fig. 9B, top), whereas expression of the S2N-S1C chimera displayed diffuse ER localization similar to STIM1 (Fig. 9B, bottom), further confirming that STIM2 preformed puncta are driven by its C terminus and not by the enhanced sensitivity of its N terminus to store depletion (6). Each STIM chimera was overexpressed in STIM1/2 Ϫ/Ϫ cells to measure SOCE in response to store depletion with thapsigargin ( Fig. 9, C and D). Stimulation of S1N-S2C-expressing cells displayed a similar release of ER Ca 2ϩ compared with S2N-S1C-expressing cells and nontransfected STIM1/2 Ϫ/Ϫ cells. However, S2N-S1C overexpression resulted in increased constitutive activity (Fig. 9, F and G). SOCE was significantly rescued with expression of either S1N-S2C or S2N-S1C chimeras. However, S2N-S1C expression supported bigger SOCE (Fig. 9,  C and D), likely the result of its enhanced constitutive activity. Interestingly, addition of 50 M 2-APB strongly inhibited SOCE mediated by the S2N-S1C chimera with less inhibitory effect on SOCE supported by the S1N-S2C chimera (Fig. 9, C-E), suggesting the STIM2 C terminus protects against 2-APBmediated inhibition. These findings are consistent with the reported constitutive activity of a similar STIM1 chimera containing the EF-SAM domain of STIM2 co-expressed with ORAI1 (45). We then determined the ability of the STIM chimeras to mediate store-independent Ca 2ϩ entry in response to 2-APB stimulation when expressed in STIM1/2 Ϫ/Ϫ cells. As expected, overexpression of STIM2 supported a sustained

Structural determinants of enhanced STIM2 sensitivity
Ca 2ϩ entry in response to 10 M 2-APB, whereas STIM1 overexpression mediated marginal Ca 2ϩ entry under the same conditions (Fig. 9, H and I). Unexpectedly, both S1N-S2C and S2N-S1C chimeras mediated minimal Ca 2ϩ entry comparable with STIM1 in response to 10 M 2-APB (Fig. 9, H and I). Similar results were observed when cells were stimulated with 50 M 2-APB with each STIM chimera displaying minimal store-independent Ca 2ϩ entry (Fig. 9, J and K). These results suggest that both the high sensitivity of the STIM2 N terminus and the communication with the flexible conformation of its C terminus are key determinants for activation of store-independent Ca 2ϩ entry in response to 2-APB, and likely in response to weak agonists or low concentrations of strong agonists.

Discussion
Recent studies have defined unique functions for STIM2 in regulating cytosolic Ca 2ϩ signals under conditions of minimal ER Ca 2ϩ depletion (18,19). Upon stimulation with low concentrations of agonist that minimally deplete intracellular stores, STIM2 has been demonstrated to recruit STIM1 into ER-PM junctions to interact with ORAI1 channels and drive SOCE under conditions where STIM1 would normally be unable to oligomerize and form puncta (18). Mechanistically, the increased flexibility of the STIM2 C-terminal OASF2 region along with higher phosphatidylinositol 4,5-biphosphate binding affinity of the C-terminal polybasic domain are suggested to account for constitutive activation and preformed puncta of STIM2 under resting conditions (19,46). The flexibility of OASF2 triggers remodeling of the STIM1 C terminus for interaction and gating of ORAI1 under high ER Ca 2ϩ concentrations, possibly through heteromeric interactions between STIM1 and STIM2 (18). Previous studies investigated the effects of high concentrations of 2-APB on STIM and ORAI using overexpression and showed that 2-APB induces avid binding of cytosolic C termini of both STIM1 and STIM2 to ORAI1 causing increased channel activity (24). Functional binding of expressed STIM1 C terminus to ORAI1 also occurred in DT40 chicken B cells lacking endogenous STIM1 and STIM2, indicating that C-terminal fragments work independently of native STIMs and that STIM C terminus is critical for 2-APB action (24). Here we show that under native levels of expression, both the N terminus and C terminus of STIM2 are necessary and sufficient for enhanced STIM2 sensitivity, suggesting that weak agonists or low concentrations of strong agonists use STIM2 to enhance the diversity of physiological Ca 2ϩ signals. Through the use of multiple CRISPR/Cas9 cell lines with STIM1 and STIM2 knocked out individually or in combination, we conclude that 2-APB-activated store-independent Ca 2ϩ entry is mediated exclusively by endogenous STIM2. This store-independent entry was significantly enhanced with STIM2 overexpression in HEK293 STIM1/2 Ϫ/Ϫ cells and augmented further when STIM2 was expressed in combination with ORAI1. Importantly, STIM2-dependent 2-APB-activated

Structural determinants of enhanced STIM2 sensitivity
Ca 2ϩ entry could not be rescued with overexpression of STIM1 alone or with STIM1 and ORAI1 overexpression. Only high concentrations of 2-APB (50 M) could induce a moderate level of store-independent Ca 2ϩ entry in STIM1/ORAI1 overexpressing cells that was ϳ 1 ⁄ 3 of that observed in cells overexpressing STIM2/ORAI1 and stimulated by the same concentration of 2-APB. By testing full-length STIM1 and STIM2 variants with mutations predicted to stabilize or disrupt CC1/CC3 interactions, we determined that increased flexibility of the STIM2 C terminus plays a significant role in response to 2-APB-mediated Ca 2ϩ entry. Introduction of a stabilization mutation (R517L) shown to restrict STIM2 C-terminal flexibility and strengthen its CC1/CC3 interactions slightly reduced Ca 2ϩ entry in response to 10 M 2-APB compared with STIM2. Importantly, introduction of a mutation into the C terminus of STIM1 (L251S) making it more flexible caused preformed STIM1 puncta and induced constitutive Ca 2ϩ entry, but could not support store-independent Ca 2ϩ entry in response to low or high concentrations of 2-APB. These data suggest that whereas increased flexibility of the STIM2 C terminus is a critical factor in the induction of store-independent Ca 2ϩ entry, it is not sufficient.
In addition to differences in CC1/CC3 interactions and differences in key residues, phenylalanine (Phe-394) in SOAR1 is a leucine in SOAR2 and glycine (Gly-379) in SOAR1 is a glutamate in SOAR2, functional differences within the STIM2 N terminus also distinguish its activity from STIM1. Substitution of three amino acids within the N-terminal EF-hand domain of STIM2 confer reduced Ca 2ϩ binding affinity, lowering the threshold of STIM2 to induce an active conformation in response to minor reductions in ER Ca 2ϩ concentrations (11). Although mutation of these residues in the STIM1 EF-hand to mimic STIM2 increases the Ca 2ϩ sensitivity of STIM1, conversion of the STIM2 EF-hand into the corresponding STIM1 residues failed to mimic the activation kinetics of STIM1 (11). Our chimeric STIM constructs demonstrated that substitution of the entirety of STIM1 N terminus with that of STIM2 failed to activate store-independent Ca 2ϩ entry in response to 2-APB stimulation. Similarly, replacement of the STIM2 N terminus with that of STIM1 reduced the response to 2-APB to levels similar to STIM1. These data suggest that enhanced sensitivity of the STIM2 N terminus in concert with the increased flexibility of its C terminus are responsible for the unique store-independent activation of STIM2.
Conventional models of SOCE and the regulation of cytosolic Ca 2ϩ signals have long placed STIM1 as the primary activator of ORAI channels in response to substantial ER depletion, whereas STIM2 is believed to act as a homeostatic

Structural determinants of enhanced STIM2 sensitivity
housekeeping sensor that maintains basal Ca 2ϩ concentrations (11,12). Unique structural features of STIM2 N and C termini dampen its interaction with ORAI channels to prevent Ca 2ϩ overload as a result of its constitutive activity in ER-PM junctions (6,11). Our data utilizing multiple STIM knockout cell lines show that depending on the cell type, native STIM2 can play a significant role in sustaining SOCE when ER Ca 2ϩ is maximally depleted. Knockout of STIM2 in HEK293 cells resulted in a minor reduction (ϳ16% inhibition) of peak SOCE compared with WT cells, whereas STIM2 knockout in HCT116 drastically reduced SOCE (ϳ56% inhibition) with no obvious compensatory effect on STIM1 protein levels in either cell line. Furthermore, we noted no differences in STIM1 and STIM2 protein expression between HEK293 and HCT116 cell lines. It is tempting to speculate that in HCT116 cells, STIM2 interacts with and remodels the C terminus of STIM1 into an active conformation despite ER Ca 2ϩ being significantly depleted. In support of this idea, knockout of STIM1 in HCT116 cells augmented store-independent Ca 2ϩ entry in response to low and high concentrations of 2-APB, suggesting that STIM1 and STIM2 might dynamically regulate the function of one another in a cell type-specific manner. Our results in HCT116 cells are in line with recent data in NIH3T3 fibroblasts and ␣T3 cells where knockout of STIM2 resulted in a 90% reduction in SOCE (47). In addition to these cell types, a multitude of studies have identified STIM2 as the predominant STIM isoform regulating SOCE in primary neurons (48 -50). STIM2mediated Ca 2ϩ entry plays a critical role in maintaining mushroom dendritic spine development in vitro and in vivo as well as regulating neuronal apoptosis in response to environmental stress (48, 50 -52). The use of 2-APB and newly characterized 2-APB analogues (53)(54)(55) will be critical in understanding the endogenous interactions between STIM1 and STIM2 and how these ER sensors positively and negatively regulate one another in various physiological systems. Our findings elucidating the differential effects of 2-APB on STIM1 and STIM2 offers readily exploitable approaches to understand the endogenous interactions between STIM1 and STIM2 and might lead to drugs that specifically distinguish STIM2-mediated from STIM1-driven (patho)physiological functions.

Generation of STIM1, STIM2, and STIM1/2 double knockout cell lines
Guide RNAs with sequences specific for STIM1 or STIM2 were designed and inserted into the BsmBI restriction site of the lentiCRISPR v2 vector (Addgene plasmid number 52961). The guide sequences used were the following: STIM1g1, 5Ј-TGATGAGCTTATCCTCACCA-3Ј; STIM2g1, 5Ј-AGATG-GTGGAATTGAAGTAG-3Ј; and STIM2g2, 5Ј-AGAAGAAG-ACAGATTTAGTC-3Ј. HEK293 cells were transfected with the cloned lentiCRISPR v2 vectors using a Nucleofector II Device (Amaxa Biosystems) and HCT116 were transfected using Lipofectamine 2000 (Invitrogen). 48 h after transfection, HEK293 and HCT116 cells were cultured with their respective media containing puromycin (2 g/ml) (Gemini Bio Products) and selected for 6 days. After puromycin selection, cells were plated at a density of one cell per well into 96-well plates to isolate individual clones. Disruption of the STIM1 or STIM2 genes was confirmed in individual clones through Sanger sequencing, Western blot analysis, and functional Ca 2ϩ imaging experiments.

Single cell Ca 2؉ imaging
HEK293 and HCT116 cells were seeded on 25-mm glass coverslips overnight, mounted in Attofluor cell chambers (Thermo Fisher Scientific), and incubated in Dulbeccos modified Eagle's or McCoy's media containing 2 M Fura-2AM (Molecular Probes) at 37°C for 30 min as described previously (56,58). Following loading with Fura-2AM, cells were washed three times and incubated for 10 min in a HEPES-buffered saline solution containing the following components: 140 mM NaCl, 1.13 mM MgCl 2 , 4.7 mM KCl, 2 mM CaCl 2 , 10 mM D-glucose, and 10 mM HEPES with pH adjusted to 7.4. Chambers were mounted onto a Nikon TS100 inverted microscope equipped with a ϫ20 Fluor objective and fluorescence images were recorded with a digital fluorescence imaging system (InCytIm2, Intracellular Imaging Inc.). Ca 2ϩ imaging was also performed on a Leica DMi8 fluorescence microscope controlled by Leica Application Suite X (Leica). Fura-2AM fluorescence was measured every 2 s by alternative excitation at 340 (F 340 ) and 380 nm (F 380 ) and emission fluorescence was collected at 510 nm. Cytosolic Ca 2ϩ concentrations are represented as the ratio of F 340 /F 380 . All Ca 2ϩ imaging experiments were performed at room temperature and traces are shown as mean Ϯ S.E. from at least three independent experiments.

Western blotting
Cell lines were harvested, washed once with chilled PBS, and lysed in pre-chilled RIPA buffer (Sigma) containing Halt Protease and Phosphatase Inhibitor Mixture (Thermo Fisher Scientific) for 10 min on ice. Samples were centrifuged at 14,000 ϫ g, at 4°C for 10 min and clarified protein supernatants were quantified with the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). 25 g of protein extract was loaded on a 4 -12% gel NuPAGE BisTris gel (Life Technologies) and transferred to a polyvinylidene difluoride membrane using a Trans-Blot Turbo Transfer System (Bio-Rad). Membranes were blocked in Odyssey Blocking Buffer (LI-COR) for 1 h at room temperature and incubated overnight at 4°C with one of the following primary antibodies: STIM1 antibody (1:2000; number 4916, Cell Signaling Technology), STIM2 (1:1000; number 4917, Cell Signaling Technology), or GAPDH (1:4000; MAB374, Sigma). Membranes were washed 3 times in 0.1% TBST and incubated for 1 h at room temperature with the following secondary antibodies: IRDye 680RD goat anti-mouse (1:10,000 LI-COR) or IRDye Structural determinants of enhanced STIM2 sensitivity 800RD donkey anti-rabbit (1:10,000 LI-COR). Following 3 washes in 0.1% TBST, membranes were imaged on an Odyssey CLx Imaging System (LI-COR). Western blotting image analysis was performed in Image Studio version 5.2 (LI-COR) and ImageJ.

Förster resonance energy transfer (FRET) measurements
Measurement of OASF FRET sensors was performed on a Leica DMI 6000B inverted automated fluorescence microscope with CFP (438 excitation/483 emission), YFP (500 excitation/ 542 emission), and FRET (438 excitation/542 emission) filter cubes. Images were acquired every 20 s with each filter cube with a ϫ40 oil objective and analyzed using Slidebook 6.0 software (Intelligent Imaging Innovations). Exposure times for each channel were: 1000 ms (CFP), 250 ms (YFP), and 1000 ms (FRET). Three channel corrected FRET was calculated with, for this formula, I DD , I AA , and I DA are the intensity of the background-subtracted CFP, YFP, and FRET images, respectively. For FRET analysis of OASF sensors, YFP-OASF-YFP and CFP-OASF-CFP were used for calculation of correction images and bleed through of CFP and YFP through the FRET filter. For analysis of OASF2 sensors, YFP-OASF2-YFP and CFP-OASF2-CFP were used for correction images. The E-FRET method to analyze 3-cube FRET images as initially described by Zal and Gascoigne (59) was calculated with the following formula.
All FRET experiments were performed by transient transfection of OASF/OASF2 constructs into HEK293 STIM1/STIM2 double knockout cells.

Statistics
All data analyses were performed with GraphPad Prism 8 (GraphPad Software) and data are presented as mean Ϯ S.E. An unpaired, nonparametric Mann-Whitney test was used for statistical comparisons between two groups and the Kruskal-Wallis test with Dunn's multiple comparisons was used for comparison between multiple groups.