Stromal Interaction Molecule 1 (STIM1) Regulates ATP-sensitive Potassium (KATP) and Store-operated Ca2+ Channels in MIN6 β-Cells*

Stromal interaction molecule 1 (STIM1) regulates store-operated Ca2+ entry (SOCE) and other ion channels either as an endoplasmic reticulum Ca2+-sensing protein or when present in the plasma membrane. However, the role of STIM1 in insulin-secreting β-cells is unresolved. We report that lowering expression of STIM1, the gene that encodes STIM1, in insulin-secreting MIN6 β-cells with RNA interference inhibits SOCE and ATP-sensitive K+ (KATP) channel activation. The effects of STIM1 knockdown were reversed by transduction of MIN6 cells with an adenovirus gene shuttle vector that expressed human STIM1. Immunoprecipitation studies revealed that STIM1 binds to nucleotide binding fold-1 (NBF1) of the sulfonylurea receptor 1 (SUR1) subunit of the KATP channel. Binding of STIM1 to SUR1 was enhanced by poly-lysine. Our data indicate that SOCE and KATP channel activity are regulated by STIM1. This suggests that STIM1 is a multifunctional signaling effector that participates in the control of membrane excitability and Ca2+ signaling events in β-cells.

The primary physiological role of pancreatic ␤-cells is to regulate metabolism by sensing changes in blood glucose concentration and secreting insulin accordingly. Interplay between glucose metabolism, closure of ATP-sensitive K ϩ (K ATP ) channels, and Ca 2ϩ signaling evoked by activation of voltage-dependent Ca 2ϩ channels (VDCCs) 2 form the molecular basis of the so-called consensus model of glucose-stimulated insulin secretion. Nevertheless, this simple model does not address the temporal and spatial diversity of Ca 2ϩ -dependent signal transduction events generated in ␤-cells consequent to stimulation with secretagogues (1).
Ca 2ϩ homeostasis and signal transduction are controlled by multiple ion transport mechanisms and organelles. Store-operated Ca 2ϩ entry (SOCE) plays a critical role regulating spatial and temporal changes in cytoplasmic Ca 2ϩ concentration ([Ca 2ϩ ] c ), endoplasmic reticulum (ER) Ca 2ϩ homeostasis and protein biosynthesis, mitochondrial function, secretion, and cell viability (2,3). The ER regulates Ca 2ϩ signaling by acting as a subcellular store for Ca 2ϩ that can be rapidly and transiently released into the cytoplasm, and the ER luminal Ca 2ϩ concentration ([Ca 2ϩ ] ER ) regulates gating of store-operated Ca 2ϩ (SOC) channels located in the plasma membrane (PM).
The presence of SOCE in ␤-cells was first reported in 1994 (4 -6). Patch clamp electrophysiology has been used to characterize some of the electrical properties of SOC current (I SOC ) in ␤-cell lines and primary ␤-cells isolated from rodent islets of Langerhans (5)(6)(7). The store-operated conductance in insulinsecreting cells was found to be an inwardly rectifying current carried by a non-selective cation (CRAN) channel rather than a highly Ca 2ϩ -selective Ca 2ϩ -release activated Ca 2ϩ (CRAC) channel (5,8). The electrical characteristics of I SOC in insulinsecreting cells are similar to mammalian transient receptor potential (trp)-related channels, a gene family that includes Ca 2ϩ store-operated channels (9,10), and several trp genes are expressed in ␤-cells (11,12). Members of the Orai family of genes (Orai1, Orai2, and Orai3) and some TRPC channels (TRPC1, TRPC2, and TRPC4) form store-operated channels either as homomeric or heteromeric complexes (13). Although Affymetrix GeneChip microarrays show that rodent ␤-cell lines as well as human islets express Orai genes (T1Dbase), the roles of the proteins encoded by these genes in ␤-cell store-operated Ca 2ϩ entry and ␤-cell stimulus-secretion coupling remain to be clarified.
Stromal interaction molecule-1 (STIM1), a single transmembrane-spanning Ca 2ϩ binding phosphoprotein located in both the ER and PM, functions as a luminal Ca 2ϩ sensor of the ER and couples changes in [Ca 2ϩ ] ER with activation of SOCE (14). STIM1 is activated by cellular stimuli and stress conditions that lower [Ca 2ϩ ] ER . After a decrease in [Ca 2ϩ ] ER , STIM1 translocates to plasma membrane-associated membrane (PAM) complexes, specialized regions where the ER and PM are closely apposed. At the PAM, STIM1 directly interacts with plasmalemmal store-operated ion channels and other protein targets that control a wide range of cellular signaling events (15,16). STIM1 is expressed in ␤-cells, but its role in ␤-cell Ca 2ϩ signaling has not been fully resolved (17). Confocal microscopy demonstrated that subsequent to decreasing [Ca 2ϩ ] ER with thapsigargin, an inhibitor of sarco(endo)plasmic reticulum calcium ATPase (SERCA), fluorescently-tagged STIM1 expressed in MIN6 ␤-cells translocates to regions near the plasma membrane (18) consistent with STIM1 regulating ␤-cell SOC channels. STIM1 translocation and formation of sub-PM puncta or clusters of STIM1 molecules can also occur in response to cAMP elevation but without activation of SOCE (19).
To better understand the role of STIM1 in ␤-cells, we utilized patch clamp electrophysiology, Ca 2ϩ imaging, and RNA interference. We demonstrate for the first time that in insulinsecreting cells STIM1 participates in the activation of storeoperated Ca 2ϩ current but also directly interacts with K ATP channels. Our data suggest that STIM1 regulates ␤-cell excitability by coupling multiple PM signaling mechanisms with changes in [Ca 2ϩ ] ER .

Results
Store-operated Ca 2ϩ Entry in MIN6 Cells-Application of carbachol, a muscarinic receptor agonist, in the absence of external Ca 2ϩ evoked a transient increase in [Ca 2ϩ ] c (Fig. 1A). After discharge of the intracellular Ca 2ϩ stores, the addition of extracellular Ca 2ϩ caused a biphasic increase in [Ca 2ϩ ] c characteristic of SOCE found in a wide range of cell types (Fig. 1A). The response of MIN6 cells with depleted stores to Ca 2ϩ reintroduction was significantly reduced (peak amplitude reduced by 11% and area under the Ca 2ϩ response curve (area-undercurve (AUC)) by 24%) by nimodipine (5 M), an inhibitor of voltage-gated Ca 2ϩ channels (Fig. 1, B and C). These data indicate that Ca 2ϩ influx through L-type voltage-gated Ca 2ϩ channels makes a small, but significant, contribution to the increase in [Ca 2ϩ ] c under these conditions. This nimodipine-sensitive influx could result from spontaneous activity of L-type channels at the resting potential (20) or from a small SOCE-induced depolarization and weak activation of VDCCs. The broad spectrum channel inhibitor, 1-[2-(4-methoxyphenyl)-2-[3-(4-methoxyphenyl) propoxy]ethyl-1H-imidazole hydrochloride (SKF 96365), is reported to inhibit SOCE in ␤ cells (8) and was also tested. MIN6 cells were plated into 96-well plates and Fura 2-loaded in Ca 2ϩ -free solution containing 1 M thapsigargin (Tg) to deplete ER stores and activate SOCE. The Fura 2 loading solution was replaced with Ca 2ϩ -free bath solution containing Tg Ϯ SKF96365. Ca 2ϩ addition to the bath solution induced SOCE that was dose-dependently inhibited by SKF96365 (Fig.  1D).
Knockdown of STIM1 Reduces SOCE in MIN6 Cells-STIM1 functions as a sensor of [Ca 2ϩ ] ER , coupling ER Ca 2ϩ store content with generation of SOCE and cytosolic Ca 2ϩ oscillations, FIGURE 1. Store-operated Ca 2؉ entry in MIN6 cells. A, MIN6 cells were loaded with Fura 2 and perifused with test solutions containing 2 mM glucose at 37°C. Cells were initially bathed in normal, 2.5 mM Ca 2ϩ extracellular solution to establish a baseline. The bath solution was then exchanged for Ca 2ϩfree solution (10 M EGTA), Ca 2ϩ -free with 250 M carbachol (CCh) followed by Ca 2ϩ -free with 10 M cyclopiazonic acid (CPA) to discharge and prevent refilling of intracellular Ca 2ϩ stores. Reintroduction of 2.5 mM Ca 2ϩ caused a rise of cytosolic [Ca 2ϩ ] ([Ca 2ϩ ] c ) representing SOCE. B, the addition of 5 M nimodipine (Nimod) before the reintroduction of extracellular Ca 2ϩ caused a small decrease in the peak amplitude. The efficacy of nimodipine to block voltage-gated Ca 2ϩ channels was confirmed by applying a pulse of 56 mM KCl solution (indicated by arrow). Whereas under control conditions KCl elicited a large rise in [Ca 2ϩ ] c , this rise was inhibited by nimodipine. C, the AUC of the SOCE response shown in B was measured for the 160-s period after Ca 2ϩ addition and was significantly reduced by nimodipine (52 control cells, 24 nimodipine-treated; *, p ϭ 0.002, ANOVA). D, MIN6 cells in a 96-well plate were loaded with Fura 2 in Ca 2ϩ -free solution containing 2 mM glucose and 1 M Tg. The cells were then washed with fresh bath solution containing Tg. Ca 2ϩ -free bath solution (Cont) or Ca 2ϩ -containing solution with Tg and SKF96365 (final, 2 mM [Ca 2ϩ ], SKF96365 concentration indicated in M) was injected at 100 s. SOCE was dose-dependently blocked by SKF96365.Data are plotted as the mean Ϯ S.E., error bars shown are larger than symbols, averaged from 12 wells for each solution and are representative of three independent experiments. but knowledge of the role of STIM1 in ␤-cell Ca 2ϩ signaling is limited. We employed a knockdown strategy using viral transduction with short-hairpin RNAs (shRNA) directed against STIM1. Cells expressing a scrambled sequence with no known target (shRNA-scr, ss) were used as controls. By qPCR screening of RNA isolated from cells expressing different shRNA constructs, we identified a construct (shRNA-STIM1, sh) that significantly reduced STIM1 mRNA expression and protein levels by 84 and 70%, respectively ( Fig. 2A). In shRNA-STIM1 cells, SOCE peak amplitude and AUC were 67 and 61%, respectively, lower than shRNA-scr control cells (Fig. 2, B-D). The specificity of shRNA construct targeting and absence of off-target effects were confirmed by transducing shRNA-STIM1 cells with an adenovirus encoding the human isoform of STIM1 (AdV-hSTIM1, av). Reconstituting STIM1 protein expression ( Fig. 2A) in this way partially restored SOCE (Fig. 2, B-D). We also used PCR with primers spanning exon 11 to test for the expression of long and short isoforms of STIM1 (21,22) and found no evidence for expression of the long isoform in MIN6 cells (data not shown).
Characterization of Inward Currents Activated by STIM1 in ␤-Cells-After confirming that SOCE occurs in MIN6 cells by Ca 2ϩ imaging, experiments to investigate ion currents were performed in the whole cell configuration using the standard bath solution and Cs ϩ pipette solution. Cells were transduced with an ER-targeted Ca 2ϩ reporter to confirm that ER stores were depleted by dialysis with the base pipette solution. Under these conditions, an inward current was activated (Fig. 3A, upper trace) that followed depletion of ER Ca 2ϩ stores (Fig. 3A, lower trace). This current reached peak amplitude within 5 min after break-in and showed slow inactivation. The inward current displayed a linear current-voltage relation that reversed at Ϫ12.7 Ϯ 2.3 mV (n ϭ 27 cells, Fig. 3B). This property is consistent with previous reports of a store-operated non-selective cation current in ␤-cells (5). A non-selective blocker of SOCE, SKF96365 (2,8), reversibly inhibited this inward current ( We determined ion selectivity of the inward current by substituting cations present in the extracellular solutions. Replacement of extracellular Na ϩ with N-methyl-D-glucamine (NMDG ϩ ) had no effect on the inward current amplitude (Fig. 4A), suggesting that the current is not carried through non-selective cation channels. Given that store depletion activates Ca 2ϩ influx, we applied a Ca 2ϩ -free extracellular solution. This Ca 2ϩ -free solution increased, rather than inhibited, this current (Fig. 4B), suggesting that the current activated under . SOCE in shRNA-STIM1 cells (sh) is inhibited relative to shRNA-scr (ss) and was partially rescued by adenoviral transduction with human STIM1 (av). Traces show the mean Ϯ S.E. The peak SOCE amplitude (C) and AUC for the first 100 s after Ca 2ϩ addition (D) were quantified for ss, sh, and av cells (28,26 and 139 cells, respectively). Both peak and AUC were significantly reduced for sh cells relative to ss cells and were significantly recovered by av (*, p ϭ 0.05; **, p Ͻ 0.01, ANOVA). these conditions is not a pure Ca 2ϩ current but is consistent with extracellular Ca 2ϩ inhibiting influx through K ϩ channels (23,24).
The base pipette solution used in these studies is K ϩ -free, and the extracellular K ϩ is 5.9 mM. Under these conditions the predicted K ϩ reversal potential is expected to approximate the observed reversal potential of the inward current. To determine whether the observed inward current could be carried by K ϩ , we tested the effects of raising extracellular K ϩ to 56 mM. This increase of extracellular K ϩ induced a large increase in current amplitude (Fig. 4C) accompanied by a positive shift in the reversal potential of the current (Fig. 4D). The reversal potential shifted from Ϫ17.6 Ϯ 2.5 mV to ϩ13.0 Ϯ 3.1 mV and recovered to Ϫ15.2 Ϯ 2.2 mV (n ϭ 7, measured using voltage ramps). This inward current was reversibly inhibited by substitution of extracellular Ca 2ϩ with Ba 2ϩ (Fig. 4E) and by 100 M tolbutamide ( Fig. 4F), suggesting involvement of K ATP channels. This hypothesis was tested by adding different concentrations of ATP to the patch pipette solutions. The addition of ATP dosedependently reduced inward current amplitudes with an IC 50 of 0.6 mM (Fig. 4G). The inhibitory effect of ATP (1 mM) was reversed by simultaneous addition of an equal concentration of ADP (Fig. 4H). These data suggest that this current is an inward current passing through K ATP channels. The inhibitory effect of SKF96365 ( Fig. 3C) is consistent with this suggestion as this compound was recently shown to reversibly inhibit K ATP channels in smooth muscle cells (25).
Several studies have described SOCE in ␤-cells (17), suggesting that a SOC current is activated in MIN6 cells after ER Ca 2ϩ store depletion. To unmask the potential contribution of a SOC current, we used a Cs ϩ pipette solution in combination with bath solutions containing either glyburide (1 M), a sulfonylurea that blocks K ATP channels, or with tetraethylammonium chloride (TEA; 140 mM, substituted for Na ϩ ) to block inward currents through K ϩ channels. In the presence of extracellular glyburide, an inward current was observed (Fig. 5A) with a reversal potential similar to that measured in the absence of glyburide (Fig. 5B). These data are inconsistent with a highly Ca 2ϩ -selective current conducted by Orai channels and suggest that if a SOC current is present, it was not adequately isolated by adding extracellular glyburide. However, in the presence of extracellular TEA, dialysis of wild-type MIN6 cells to deplete ER stores generated a small inward current with a mean current density of 0.5 Ϯ 0.1 pA/pF (n ϭ 7 cells); this current was inhibited by exposing the cells to a Ca 2ϩ -free bath solution (Fig. 5C). This small Ca 2ϩ current is likely to represent activation of SOC channels in ␤-cells. Under these recording conditions, activation of the store-independent arachidonic acid-regulated Ca 2ϩ (ARC)-channel current present in ␤-cells (26) is not expected to occur.
The store-operated Ca 2ϩ current ( Fig. 5C) in shRNA-scr MIN6 cells was 0.2 Ϯ 0.02 pA/pF (n ϭ 11 cells). Knockdown of STIM1 expression in shRNA-STIM1 cells suppressed this current below the limit for reliable measurement. These data are consistent with the expression of a STIM1-regulated store-operated Ca 2ϩ -influx pathway in MIN6 cells similar to that described in other types of cells (27).
STIM1 Regulates K ATP Currents in MIN6 Cells-Ca 2ϩ imaging and electrophysiology data shows that knockdown of STIM1 using shRNA results in the predicted reduction of SOCE in MIN6 cells. However, STIM1 has previously been shown to interact with other target proteins including an inhibitory effect on Ca V 1.2 L-type Ca 2ϩ channels (28 -30) that is important for triggering insulin secretion (31,32). To test for the potential effects of STIM1 on alternative targets important for regulating insulin secretion from ␤-cells, we measured membrane currents in cells stably expressing shRNA-STIM1 or control shRNA-scr. Knockdown of STIM1 reduced the amplitude of the K ATP current by 54% (11.1 Ϯ 1.6 pA/pF, n ϭ 18 cells and 5.1 Ϯ 1.1 pA/pF, n ϭ 30 cells) for shRNA-scr (ss) and shRNA-STIM1 cells (sh), respectively (Fig. 6A). Reconstitution Under these recording conditions, with normal bath solution, an inward current developed, and the effects of ion substitution were tested. A, the current was not inhibited by substituting extracellular Na ϩ with NMDG ϩ (NMG, application indicated by filled bar). B, the inward current amplitude increased upon removal of extracellular Ca 2ϩ (Ca-free, filled bar). C, raising extracellular K ϩ from 5.9 to 56 mM (56 K, filled bar) produced an increase in current amplitude (Ϫ30.9 Ϯ 9.1 pA/pF at normal extracellular K ϩ and Ϫ247.2 Ϯ 75.0 pA/pF at 56 mM K ϩ , n ϭ 7). D, increasing K ϩ produced a positive shift of the reversal potential for the current. In the cell illustrated, the reversal potential shifted from Ϫ16.1 mV to ϩ15.9 mV and recovered to Ϫ14.3 mV, the mean change was from Ϫ17.6 Ϯ 2.5 mV at normal (5.9 mM K ϩ ) to ϩ13.0 Ϯ 3.1 mV at 56 mM K ϩ and Ϫ15.2 Ϯ 2.2 after returning to normal K ϩ . I m , inward current. E, substitution of extracellular Ca 2ϩ for Ba 2ϩ also inhibited the inward current. Current inhibition caused by a 10-s pulse of Ba 2ϩ was 97.1 Ϯ 1.2% (n ϭ 7). F, the inward current was reversibly inhibited by 100 M tolbutamide (Tolbut). G, increasing concentrations of ATP in the pipette solution inhibited the inward current (I m ). Data points were curvefitted indicating an IC 50 of 0.6 mM (data from 9 -13 cells at each concentration). H, inhibition of inward current (I m ) by 1 mM ATP was reversed by 1 mM ADP (C, control (no ATP), n ϭ 13 cells; ATP, n ϭ 9 cells; ATPϩADP, n ϭ 9 cells).
STIM1 Protein Interacts with the Sulfonylurea Receptor 1 (SUR1) Subunit of K ATP Channels-To further investigate the potential role of STIM1 in regulating K ATP channels, we performed immunoprecipitation studies to determine whether there was a physical interaction between STIM1 and the K ATP channel subunits SUR1 and Kir6.2. For these studies mouse islet protein extracts or lysates from HEK293T cells transfected with human STIM1(N131Q/N171Q) and FLAG-tagged SUR1 with or without Kir6.2 were used. The Asn-131/171 mutations in STIM1 prevent glycosylation and cell surface expression (33). Cell lysates were subject to immunoprecipitation (IP) using antibodies against SUR1, FLAG, or STIM1 with pulldown using protein A-agarose beads followed by Western immunoblotting (IB). All three lysates showed pulldown of STIM1 by the anti-SUR1 antibody (Fig. 7A). These data suggest an interaction between STIM1 and SUR1 as pulldown occurred in the absence of Kir6.2 in the HEK cell lysates (Fig. 7, C and D). This experiment was reversed with IP by anti-STIM1 and IB using anti-SUR1. Pulldown of SUR1 was observed in all three samples and was enhanced by poly-lysine (50 g/ml) addition during the IP step (Fig. 7B). Incubation of cell lysates with anti-FLAG antibody is shown to pull down STIM1 in cells expressing FLAG-SUR1 but not in cells that did not express FLAG-SUR1 (Fig. 7C). Similarly, cell lysates treated with protein A-agarose beads alone did not pull down STIM1. These data demonstrate that the pulldown of STIM1 is not due to direct interaction of and nimodipine (10 M) were added to the bath solution to block K ATP and L-type VDCCs, respectively, in cells dialyzed with Cs ϩ pipette solution containing 5 mM EGTA. Under these conditions a small inward current developed. B, measurement of the reversal potential of the current shown in A indicates that this current is not Ca 2ϩ -selective. I m , membrane current; V m , membrane potential. C, using an extracellular solution containing TEA (substituted for Na ϩ ) to block inward K ϩ currents through K ATP channels, a small inward current carried by Ca 2ϩ was revealed. This current was inhibited by transient puffer applications of Ca 2ϩ -free bath solution, indicated by filled bars. The current illustrated is representative of seven cells with a mean current density of 0.5 Ϯ 0.1 pA/pF.  Fig. 3, by ϳ50% relative to shRNA-scr (ss) controls (*, p ϭ 0.014; ss, n ϭ 18 cells; sh, n ϭ 24 cells), and this effect was reversed by adenoviral transduction of sh cells with human STIM1 (sh, n ϭ 6 cells; sh ϩ av, n ϭ 8 cells; **, p Ͻ 0.001, ANOVA). B and C, STIM1 knockdown had no effect on the amplitude of voltage-dependent Ca 2ϩ currents measured using the same solutions with voltage steps from Ϫ60 mV to 0 mV ( ss, n ϭ 7 cells; sh, n ϭ 8 cells, not significantly different, ANOVA). A representative example of a record is shown in B, and quantification of peak current data for ss and sh cells is shown in C. D and E, exocytosis measured by capacitance tracking in cells dialyzed with 1.5 M Ca 2ϩ was also unaffected by STIM1 knockdown. A representative example of a capacitance record is shown in D, and quantification of the data for ss and sh cells shown in E (ss, n ϭ 10 cells; sh, n ϭ 8 cells; ns, not significantly different, ANOVA). The rate of exocytosis was determined by linear fitting of the data using Origin software. For the cell shown in D, the linear fit is illustrated and gave a y intercept of 2.05 pF, used to normalize data to cell size, and the slope was 0.76 fF/pF/s. C m , whole cell capacitance.
the FLAG antibody or of the beads with STIM1. Pulldown of STIM1 using anti-FLAG IP was observed in either the presence or absence of Kir6.2 (Figs. 7, A-D), suggesting that STIM1 interacts with SUR1.
Poly-lysine is known to reverse the stimulatory action of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) on K ATP channels (34) and enhance sulfonylurea binding to SUR1 (35). PIP 2 also inhibits the binding of syntaxin-1A to SUR1, reversing syn-FIGURE 7. Interaction of STIM1 and K ATP . A, IP studies using mouse islet (mIslet) extracts or HEK293T cell extracts expressing STIM1 and FLAG-SUR1 in the presence or absence of Kir6.2. IP was performed using anti-SUR1 antibody followed by IB with anti-STIM1.The SUR1 antibody pulled down STIM1 in all three samples. The input protein (IB only with anti-STIM1) is shown in the right three lanes. The band indicated as HC represents the heavy chain of the IP antibody, detected by the secondary antibody, and is only seen in the three IP samples. Co-immunoprecipitation was observed in two independent experiments using mouse islet or HEK cell proteins performing IP with anti-STIM1 and IB with anti-SUR1. B, IP was performed using anti-STIM1 and IB with anti-SUR1 using the same protein samples as in A, representative of two independent experiments with this protocol. The STIM1 antibody pulled down SUR1 in all three samples, and pulldown was enhanced by adding 50 g/ml poly-lysine to the samples during IP. The SUR1 antibody exhibited a secondary band of slightly higher molecular weight than the heavy chain band, and subsequent experiments were performed using anti-FLAG to avoid any confounding effects of this band. LC represents the light chain of the IP antibody, and In is the mouse islet input protein. C, control IP studies using HEK293T cells expressing STIM1 in the presence or absence of FLAG-SUR1. IP performed with protein A-agarose beads (Beads) alone with samples expressing STIM1 and FLAG-SUR1 failed to pull down STIM1. IP with the FLAG antibody in cells expressing STIM1 but not FLAG-SUR1 also failed to pull down STIM1. IP using the FLAG antibody did, however, pull down STIM1 in cells expressing both STIM1 and FLAG-SUR1. STIM1 was detected by IB with anti-STIM1 along with nonspecific bands for the heavy-chain (HC) and light-chain (LC) of the FLAG antibody in the two FLAG-IP lanes but not the beads-only lane. IP of STIM1 using the FLAG antibody was observed in six experiments using four different protein samples. D, IP with the FLAG antibody pulled down STIM1 in the presence or absence of Kir6.2 under control conditions, and this pulldown was enhanced by the addition of 50 g/ml poly-lysine (PolyLys). The image shown is representative of three independent experiments, and the increase in band density with poly-lysine was 27 Ϯ 3% (n ϭ 9, range 17 Ϫ44%) relative to control for cells expressing STIM1 and FLAG-SUR1 and 11 Ϯ 1% (n ϭ 6, range 8 -16%) for cells expressing STIM1, FLAG-SUR1, and Kir6.2. Densities of STIM1 IB bands were significantly higher for poly-lysine-treated extracts relative to controls for these samples with STIM1 and FLAG-SUR1 (paired t test, p ϭ 0.003) and also for samples with STIM1, FLAG-SUR1, and Kir6.2 (paired t test, p Ͻ 0.001). E, the STIM1 band density was 97 Ϯ 2% (n ϭ 6, range 92-102%, not significantly different, paired t test) relative to control for cells expressing STIM1 and FLAG-SUR1 and 94 Ϯ 5% (n ϭ 6, range 83-105%, not significantly different, paired t test) for cells expressing STIM1, FLAG-SUR1, and Kir6.2. F, the FLAG band density was significantly lower with poly-lysine (91 Ϯ 6%, n ϭ 10, range 75-133%, p ϭ 0.04, three independent experiments, paired t test) relative to control for cells expressing STIM1 and FLAG-SUR1 and 83 Ϯ 5% (n ϭ 7, range 74 -99%, p ϭ 0.03, two independent experiments, paired t test) for cells expressing STIM1, FLAG-SUR1, and Kir6.2.

STIM1 Activates K ATP and SOC Channels in ␤-Cells
FEBRUARY 10, 2017 • VOLUME 292 • NUMBER 6 taxin-1A mediated K ATP channel inhibition (36). These studies led us to investigate whether poly-lysine might influence the interaction of STIM1 with SUR1 either through the effects on PIP 2 or by competing with the lysine-rich C terminus of STIM1 for binding. The interaction of STIM1 and FLAG-SUR1 was enhanced by the addition of 50 g/ml poly-lysine to the IP reaction (Fig. 7D). This effect was more pronounced in studies with STIM1 and FLAG-SUR1 without Kir6.2 than in studies with Kir6.2 (3 independent experiments, mean band density 1.27 Ϯ 0.03 higher with poly-lysine for samples expressing STIM1 and FLAG-SUR1 and 1.11 Ϯ 0.01 higher for samples with STIM1, FLAG-SUR1, and Kir6.2). This enhancement is similar to the reported effect of poly-lysine to increase sulfonylurea binding to SUR1 by ϳ30% (35). To determine whether the effect of poly-lysine was due to enhanced binding of the IB antibody to its target, we performed control Western blots using anti-STIM1 (Fig. 7E) or anti-FLAG (Fig. 7F) in the presence or absence of poly-lysine. Poly-lysine had no significant effect on anti-STIM1 binding and slightly, but significantly, decreased anti-FLAG binding. These data suggest that the enhanced IP signal in the presence of poly-lysine is due to a change in binding of STIM1 and SUR1.
To further investigate the interaction of STIM1 with SUR1, we transfected cells with STIM1 and a plasmid encoding the FLAG-tagged nucleotide binding fold-1 (FLAG-NBF1) region of SUR1. NBF1 has previously been shown to play a role in binding EPAC2 (37) and syntaxin-1A (38,39). IP studies revealed that STIM1 and FLAG-NBF1 co-immunoprecipitated, suggesting that NBF1 forms at least one interaction site between STIM1 and SUR1 proteins (Fig. 8).
To determine whether shRNA-STIM1 has an off-target effect on expression of the K ATP channel subunits SUR1 and inwardly rectifying K ϩ channel 6.2 (Kir6.2), we performed qPCR. SUR1 and Kir6.2 mRNA levels were not different in cells expressing shRNA-scr or shRNA-STIM1. In three independent experiments performed with triplicate samples in each, Kir6.2 expression in shRNA-STIM1 cells ranged from 92 to 117% of shRNA-scr controls, and SUR1 expression was 104 -157% of controls. These data indicate that the reduction in inward cur-rent amplitude in shRNA-STIM1 cells was not due to decreased expression of K ATP channel subunits and are consistent with an unexpected regulatory role for STIM1 in the control of K ATP channel activity.

Discussion
Previous work in MIN6 cells as well as mouse and human primary ␤-cells demonstrated translocation and accumulation of fluorescently labeled STIM1 to punctae in subplasmalemmal ER after ER Ca 2ϩ store depletion with thapsigargin or cyclopiazonic acid, inhibitors of SERCA (sarco(endo)plasmic reticulum calcium ATPase) pumps (18,19). The role of endogenous STIM1 in ␤-cell signaling was not established in those studies. Here, we demonstrate that STIM1 regulates SOCE in MIN6 cells. We also discovered that STIM1 interacts with SUR1, a subunit of K ATP channels, and functions in K ATP channel activation. Binding of STIM1 to SUR1 was detected in transfected HEK293T cells as well as mouse islets of Langerhans. Our data indicate that STIM1 interacts with multiple proteins in ␤-cells that contribute to the regulation of ␤-cell Ca 2ϩ signaling dynamics and membrane excitability. These findings suggest that STIM1 is a regulator of multiple signaling events in insulinsecreting cells.
It is known that SOCE in non-␤-cells is regulated by direct interaction of STIM1 with the Ca 2ϩ channel protein Orai1 and other regulatory proteins (27). Whether endogenously expressed Orai1 and STIM1 functionally interact and form the molecular basis of SOCE in ␤-cells has not been established. In view of previous reports that store depletion activates a nonselective cation current in ␤-cells (8), it seems unlikely that Orai1 would be the sole channel-forming subunit as these channels are highly Ca 2ϩ selective when activated by STIM1 (40). Non-selective store-operated currents can be formed by an Orai1-TRPC1 complex regulated by STIM1 (41)(42)(43)(44)(45). An alternative possibility is that STIM1-Orai1 forms the SOCE mechanism and that the previously reported non-selective current represents the secondary activation of Ca 2ϩ -sensitive Trp channels. Ca 2ϩ -activated TRPM4 and TRPM5 channels have been shown to play a role in regulating insulin secretion (46 -49). Expression of a Ca 2ϩ -selective SOCE mechanism distinct from previously reported non-selective cation currents would be consistent with the small currents we observe using whole cell recording.
An unexpected discovery from our experiments was that knockdown of STIM1 using shRNA reduced K ATP channel activation. We considered the possibility that this represented an "off target" effect of the shRNA. However, reconstitution of K ATP channel activation in shRNA-STIM1-expressing cells transduced with an adenovirus construct that encoded human STIM1 excludes a nonspecific effect of shRNA-STIM1. This suggests that STIM1 plays a role in regulating K ATP channel activity. Quantitative PCR assays indicated that shRNA-STIM1 did not affect SUR1 or Kir6.2 expression, therefore, indicating it unlikely that STIM1 participates in K ATP channel subunit biosynthesis. Nevertheless, our studies raise the prospect that loss of STIM1 expression either directly impacts K ATP channel gating or indirectly modifies K ATP channel activation via interaction with K ATP channel regulatory proteins. A physical interac-  anti-STIM1 (B). The data shown are representative of three independent IP experiments using four protein samples from different transfections. A, two protein samples from different transfections were collected and subject to IP using anti-FLAG followed by IB using anti-STIM1. The two left lanes show the results of IP, whereas the two right lanes show the corresponding input protein. The bands marked HC and LC that appear only in the IP lanes represent the heavy and light chains of the IP antibody. B, the same two protein samples as used in A were subjected to IP using anti-STIM1 followed by IB with anti-FLAG. IP lanes are shown on the left side of the panel, whereas the input protein is shown on the right. IP with anti-STIM1 pulled down FLAG-NBF1 and additional bands for the heavy and light chains of the IP antibody can be seen only in the IP lanes.

STIM1 Activates K ATP and SOC Channels in ␤-Cells
tion between STIM1 and SUR1 mediated through NBF1 is suggested by our immunoprecipitation data, but it is not yet clear whether this is a direct interaction or indirect through an intermediary protein.
At least two mechanisms for direct interaction of STIM1 and SUR1 are possible based on the distinct mechanisms by which STIM1 regulates Orai and TRPC1. Orai1 is activated by interaction of the STIM-Orai activation region (SOAR) of STIM1 with cytosolic N-and C-terminal domains of Orai1 (50). In contrast, STIM1 activates TRPC1 through an electrostatic interaction of a KK amino acid pair in STIM1 with a DD pair in TRPC1 (51). In silico analysis of human SUR1 reveals a region with homology to the extended transmembrane Orai1 N-terminal (ETON) domain of Orai1 that interacts with STIM1 and also four DD pairs, including two pairs within NBF1. NBF1 is known to be important for MgADP activation of K ATP (52). Additional experiments will be necessary to establish whether either of these sites plays a role in the molecular mechanism for STIM1-regulated K ATP activity.
A number of STIM1 binding partners have been identified in addition to Orai and TRP channels (53,54). Whether any of these alternative binding partners might contribute to an indirect effect on K ATP channel regulation by STIM1 is unknown. One possibility is that actin binds STIM1 (53) and the actin cytoskeleton regulates K ATP channels in ␤-cells (55,56). Leptin is a hormone that activates K ATP channels (55,57), an effect mediated by disruption of actin filaments (55). Recent work has shown that Ca 2ϩ influx through store-operated channels promotes actin depolymerization (58), raising the possibility that co-localization of STIM1 with K ATP channels might promote local actin depolymerization and channel activation. Indeed, studies of STIM1 translocation in ␤-cells indicate aggregation of STIM1 punctae in actin-poor regions of the membrane (18). Thus, the effect of STIM1 on K ATP might be an indirect effect of localized Ca 2ϩ influx through Orai channels causing cytoskeletal actin disruption.
Whether the regulation of K ATP activity is mediated by STIM1 in the ER, similar to the regulation of Orai1 and SOCE, or by plasma membrane-resident STIM1, similar to the ARC (arachidonic acid-regulated Ca 2ϩ ) channel, is unknown at present. However, our immunoprecipitation studies indicate that a mutated STIM1, which is not glycosylated and inserted in the PM (33), and SUR1, expressed in the absence of Kir6.2, interact. These data suggest that STIM1 and SUR1 interact within the ER but do not exclude interplay of PM STIM1 with SUR1.
It is interesting to note that whereas reconstitution of the expression of STIM1 in MIN6 cells using human STIM1 restored K ATP activity to above control levels, SOCE was only partially restored. Western blot analysis revealed that our rescue studies increased STIM1 protein levels above that seen in control cells (Fig. 6). A recent report indicates that high STIM1 expression can trap Orai1 intracellularly and prevent its trafficking into the plasma membrane (59). Our observation of partial rescue of SOCE may be attributable to an unfavorable ratio of STIM1 to Orai1 given that Adv-hSTIM1 raised expression of STIM1 protein to 7-12-fold higher than endogenous levels. K ATP channels also undergo dynamic trafficking to the PM (60 -63). Our data indicating that overexpression of STIM1 fully restored K ATP activity but only partially restored SOCE might indicate that STIM1 does not affect K ATP translocation in the same way as Orai1, if the trafficking trap mechanism applies in our studies. However, whether STIM1 influences K ATP trafficking remains to be determined.
The K ATP channel is well established as a key regulator of ␤-cell electrical excitability in response to nutritional status and hormonal inputs (57,64,65). Our data indicate that STIM1, through either direct or indirect interaction, regulates the activity of K ATP channels in addition to regulating SOCE and plays a role in regulating ␤-cell excitability through these two processes. We propose a working model that incorporates a physiological role for STIM1 in regulating K ATP channel activity and thereby ␤-cell membrane potential and Ca 2ϩ oscillations. After ␤-cells are exposed to a high concentration of extracellular glucose, an initial fall in cytosolic Ca 2ϩ concentration occurs (termed Phase 0) (66,67). This phase of the Ca 2ϩ response to glucose stimulation precedes ␤-cell depolarization and is caused by glucose-induced Ca 2ϩ uptake into ER Ca 2ϩ stores (67). Estimates of the resting [Ca 2ϩ ] ER in ␤-cells are 250 -500 M (68 -70), a similar range to estimates for the K D of Ca 2ϩ binding to the EF-SAM domain of STIM1 (71,72). This suggests that store filling induced by glucose elevation might decrease STIM1 activity and potentially play a role in regulating K ATP . An increase in [Ca 2ϩ ] ER will induce translocation of STIM1 away from the plasma membrane to inactivate SOCE. A second possibility is that the rise in intracellular ATP will cause displacement of STIM1 from binding to SUR1, analogous to the effects of ATP on syntaxin binding (39). Displacement of STIM1 from SUR1 by ATP would contribute to the inhibition of K ATP channels after a rise in blood glucose.
Our new observation that STIM1 knockdown inhibits K ATP activation suggests that STIM1 has a positive effect on channel activity. Therefore, the translocation of STIM1 away from the plasma membrane or displacement of STIM1 from SUR1 might be predicted to reduce K ATP activity and could contribute to channel inhibition after a rise in blood glucose. The importance of this mechanism relative to inhibition caused by the increase in the ATP/ADP ratio is unknown. However, STIM1-mediated changes in SOCE and K ATP activity could contribute to the initiation of cell depolarization and [Ca 2ϩ ] c oscillations. In contrast, release of intracellular stores will cause translocation of STIM1 to PAM regions where it can interact with and activate Orai1 to refill stores and perhaps also interact with and activate K ATP .
It has been demonstrated that SOCE activates type 8 adenylyl cyclase in both HEK 293 cells and MIN6 cells (73). This effect could lead to elevation of intracellular cAMP that is an important regulator of ␤-cell electrical activity and secretion mediated by protein kinase A and Epac2 (64,74). A role for SOCE in regulating [Ca 2ϩ ] c oscillations in ␤-cells has been proposed as a consequence of ATP consumption and regulation of K ATP activity (75). Metabolic oscillations and slow changes in K ATP activity have been proposed to be important for glucose-induced oscillations (76). It is possible that slow changes in K ATP activity as a consequence of store depletion and refilling and trafficking of STIM1 might contribute to normal bursting activity in islets.

STIM1 Activates K ATP and SOC Channels in ␤-Cells
FEBRUARY 10, 2017 • VOLUME 292 • NUMBER 6 In summary, we report that STIM1 is a multifunctional regulator of signaling dynamics in ␤-cells. In addition to defining the role of STIM1 as a regulator of SOCE in ␤-cells, our data revealed that STIM1 participates in K ATP channel activation through an interaction with SUR1. This suggests a novel pathway for the control of ␤-cell excitability and function by interplay between ER Ca 2ϩ stores and K ATP channels.
[Ca 2ϩ ] c measurement in some studies was performed using a FlexStation 3 plate reader in 96-well plate format using solutions and wavelengths for recording as described above. Test solutions were injected into individual wells containing MIN6 cells at 60 -80% confluence. Experimental parameters and data acquisition were controlled using SoftMax Pro 5.4.6 software (Molecular Devices).
Electrophysiology-Whole cell recordings of membrane currents were made at room temperature (21-24°C) using an EPC-9 amplifier controlled by PatchMaster software (HEKA Electronik, Lambrecht/Pfalz, Germany). Electrodes were pulled from patch clamp glass capillaries (catalog no. G85150T-4, Warner Instruments, Hamden, CT) using a Sutter P97 puller. MIN6 cells were bathed in KRBH. Isosmotic substitution of NaCl (NMDG-Cl or KCl) was used in some experiments, or CaCl 2 was omitted (Ca ϩ -free, 10 M EGTA) or substituted with BaCl 2 (in this Ba 2ϩ solution, MgSO 4 was replaced with MgCl 2 ). For some experiments a TEA bath solution comprised 140 mM TEA-Cl, 10 mM CaCl 2 , 1.2 mM MgCl 2 , and 10 mM HEPES (pH 7.4). Glucose was added to these solutions as indicated. Patch pipettes were filled with a base solution containing 90 mM Cs 2 SO 4 , 10 mM NaCl, 1 mM MgCl 2 , 5 mM HEPES, and 0.2 mM EGTA, pH adjusted to 7.4 with CsOH. ATP, ADP, and GTP were added to the pipette solution as indicated. Cells were held at Ϫ60 mV for current recording, and voltage ramps were applied to measure reversal potentials. PatchMaster software compensated junction potentials and was used to generate voltage ramps at 1 V/s, the software was also used to measure cell capacitance to normalize current amplitudes to cell size. Whole cell capacitance measurements of Ca 2ϩ -dependent exocytosis were made using the LockIn module of PatchMaster with an applied sine wave of 10 mV amplitude at 1 kHz and a pipette solution containing (in mM) 125 cesium glutamate, 10 NaCl, 10 KCl, 1 MgCl 2 , 10 EGTA, 11 CaCl 2 , 3 MgATP, 0.2 LiGTP, and 5 HEPES (pH 7.4). Extracellular test solutions were applied by pressure ejection from glass micro-capillaries using a Pico-Spritzer III (Parker Instrumentation). Experiments with fluorescence imaging used a PTI DeltaRam monochromator light source controlled with MetaFluor software. For FRET experiments, emission images at 485 and 535 nm were captured simultaneously using a DualView beam splitter and MetaFluor software.
Quantitative Real-time PCR-Total RNA was isolated and purified from two to four replicate plates of each cell type using an RNeasy Mini kit (Qiagen) according to the manufacturer's instructions. RNA (1 g) was reverse-transcribed to cDNA using oligo(dT 15 ) primers according to the manufacturer's instructions (Promega). Primer sequences for S18 were 5Ј-GCC-ATCACTGCCATTAAGGG and 5Ј-CCAGTCTGGGATCT-TGTACTG and for STIM1 were 5Ј-GAAGCAAAATGCAGA-GAGGC and 5Ј-CATCATCCAGGGAAGAGCTG. Primers for Kir6.2 were 5Ј-GTGTTCACCACGCTGGTGGAC and 5Ј-CAGGTCACCGTGGGCGAAGG. Primers for SUR1 were 5Ј-GGACAGAAGATCGGGATCTGC and 5Ј-GAAGGAGGAC-TTCCCACTGCC. Each primer was used at a concentration of 500 nM in a 25-l reaction volume containing Brilliant II SYBR Green qPCR master mix (Agilent Technologies). Quantitative real-time PCR was completed in a Stratagene MX3000P realtime PCR thermal cycler under the following conditions: initial hold at 95°C for 10 min followed by 40 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s. Fluorescence intensity data were collected during each cycle at 72°C. Amplicon dissociation was initiated immediately after the final PCR extension, with a 30-s hold at 72°C followed by increasing the temperature to 95°C with 0.25°C steps. One cDNA was produced from each RNA sample, and each sample was tested in duplicate by qPCR. C q values were averaged for duplicates, and relative quantification calculations were completed as described by Pfaffl (79).
Inhibition of STIM1 Expression Using shRNA-Cells in which STIM1 expression was transiently or stably deleted were used. For transient suppression of STIM1, MIN6 cells were transfected with a plasmid vector that expressed small interfering RNA (siRNA) targeted to STIM1. MIN6 cells stably expressing short hairpin RNA targeted to STIM1 (shRNA-STIM1) or shRNA containing a scrambled sequence that targeted no known genes (shRNA-scr) were produced by transduction with lentiviral particles (Santa Cruz Biotechnology). Stably transduced cells were maintained in DMEM-H medium containing 2 g/ml puromycin. In some experiments, STIM1 expression was reconstituted by transducing cells with an adenoviral vector to express the human isoform of STIM1 (AdV-shRNA-hS-TIM1) (ViraQuest Inc.).
For immunoprecipitation, 100 g of protein lysate was incubated for 16 h at 4°C with antibodies (1-5 g) as indicated. Then the antigen-antibody complex was immobilized on protein A-agarose beads (20 l of 50% beads aqueous slurry, Thermo Scientific) for 3 h at 4°C. The beads were washed 5 times with 250 l of CHAPS lysis buffer (50 mM Tris-Base, 150 mM NaCl, 1 mM EDTA, 1% CHAPS (pH 8.0)) containing 10% ␤-mercaptoethanol and pelleted by centrifugation at 13,000 rpm for 30 s at 4°C. The beads were resuspended with 20 l of 2ϫ SDS sample buffer (final concentrations: 25 mM Tris, 192 mM glycine, 1% SDS (pH 8.3)), and 20 l of radioimmune precipitation lysis buffer. The samples were then centrifuged for 60 s at 13000 rpm, and the supernatant was used for Western blotting. Primary antibodies used were anti-SUR1 (Abcam catalog no. S289-16, Lot # GR-103710-18), anti-FLAG, clone M2 (Sigma catalog no. F1804, lot # SLBN5629V), and anti-STIM1 (as described for the Western blots above). Antibody specificity was assessed in HEK293T cells by comparing Western blots of control cells with cells transfected with the appropriate plasmid to induce protein expression.
The immunoprecipitated protein samples (ϳ30 g of protein) and molecular weight markers (Precision Plus All Blue Protein Standard, Bio-Rad catalog no. 161-0373) were then resolved by SDS-PAGE (4 -20% gradient protean GTX mini gels, Bio-Rad) and wet-transferred to Immobilon-P PVDF membranes (Millipore) using a buffer containing 25 mM Trisbase, 192 mM glycine, and 20% methanol (pH 8.3). The blots were washed once with TBST and incubated for 45 min at room temperature with blocking buffer (TBST with 5% blotting grade nonfat skim milk (Bio-Rad); membranes were incubated with specific antibodies (1:1,000 dilution in blocking buffer) for 16 h at 4°C followed by 3 washes with TBST and incubation with HRP-coupled anti-mouse secondary antibody (Abcam catalog no. ab6728, lot # GR152005-0; 1:10,000 dilution) for 1 h at room temperature. After 3 washes with TBST, membranes were incubated with SuperSignal West Pico Chemiluminescence Substrate kit (Thermo Scientific) for 5 min, and protein bands were detected using a Bio-Rad Gel Documentation system with QuantityOne software. For quantification of band density, background-corrected measurements were made on images with no saturated pixels from three different exposure times to test for linearity.
Statistical Analysis-Student's t test or ANOVA was used for intergroup comparisons (p Ͻ 0.05 was considered statistically significant) as detailed in the figure legends. Data traces are plotted as the mean Ϯ S.E. Box and whisker plots show mean (solid circle), median (dashed line), 25 and 75% values (box), and minimum/maximum values (whiskers).