cAMP Induces Stromal Interaction Molecule 1 (STIM1) Puncta but neither Orai1 Protein Clustering nor Store-operated Ca2+ Entry (SOCE) in Islet Cells*

Background: Regulation of the ER Ca2+ sensor STIM1 and its association with the plasma membrane channel Orai1 to activate store-operated Ca2+ entry (SOCE) remains incompletely understood. Results: cAMP induced plasma membrane translocation of STIM1 in islet cells without concomitant SOCE activation. Conclusion: STIM1 translocation alone is insufficient to activate SOCE. Significance: This study describes novel interaction between the components of cAMP and Ca2+ signaling cascades. The events leading to the activation of store-operated Ca2+ entry (SOCE) involve Ca2+ depletion of the endoplasmic reticulum (ER) resulting in translocation of the transmembrane Ca2+ sensor protein, stromal interaction molecule 1 (STIM1), to the junctions between ER and the plasma membrane where it binds to the Ca2+ channel protein Orai1 to activate Ca2+ influx. Using confocal and total internal reflection fluorescence microscopy, we studied redistribution kinetics of fluorescence-tagged STIM1 and Orai1 as well as SOCE in insulin-releasing β-cells and glucagon-secreting α-cells within intact mouse and human pancreatic islets. ER Ca2+ depletion triggered accumulation of STIM1 puncta in the subplasmalemmal ER where they co-clustered with Orai1 in the plasma membrane and activated SOCE. Glucose, which promotes Ca2+ store filling and inhibits SOCE, stimulated retranslocation of STIM1 to the bulk ER. This effect was evident at much lower glucose concentrations in α- than in β-cells consistent with involvement of SOCE in the regulation of glucagon secretion. Epinephrine stimulated subplasmalemmal translocation of STIM1 in α-cells and retranslocation in β-cells involving raising and lowering of cAMP, respectively. The cAMP effect was mediated both by protein kinase A and exchange protein directly activated by cAMP. However, the cAMP-induced STIM1 puncta did not co-cluster with Orai1, and there was no activation of SOCE. STIM1 translocation can consequently occur independently of Orai1 clustering and SOCE.

through the plasma membrane (PM) is activated by release of Ca 2ϩ from the endoplasmic reticulum (ER) in response to Ca 2ϩ -mobilizing messengers like inositol 1,4,5-trisphosphate (IP 3 ). Some of the released Ca 2ϩ is inevitably transported out of the cell, and the SOCE compensates for this loss and ensures proper Ca 2ϩ refilling of the ER. The molecular mechanisms underlying the SOCE remained elusive for almost 2 decades until it was discovered that the stromal interacting molecule 1 (STIM1) has a role in store-operated entry (2,3). STIM1 and the structurally related STIM2 are single pass transmembrane molecules in the ER, and STIM proteins are also present in the PM (4). Both molecules contain Ca 2ϩ -sensing EF-hand motifs in their N termini facing the ER lumen and are believed to sense the Ca 2ϩ depletion that activates SOCE (3,5,6). Another molecule in the PM named Orai was identified as an important player in the store-operated mechanism (7)(8)(9). There are three isoforms of Orai (Orai1-3), and evidence has accumulated that Orai is the pore-forming unit of the store-operated channel (10 -14). Emptying of Ca 2ϩ from the ER results in dissociation of the ion from STIM1, which rapidly moves and aggregates in the subplasmalemmal ER where it forms distinct puncta (3,12,15,16), and this is the site where STIM interacts with Orai to activate SOCE (11,17,18).
The store-operated pathway is also present in excitable cells like the insulin-releasing ␤and glucagon-secreting ␣-cells. Although these cells have more potent voltage-operated routes for Ca 2ϩ entry, SOCE has been found to be important for different cellular processes. However, even maximally activated SOCE has modest effects on the cytoplasmic Ca 2ϩ concentration ([Ca 2ϩ ] i ) in ␤- (19 -21) and ␣-cells (22) and has mostly been attributed a functional role based on the depolarizing effect (22)(23)(24). Whereas Ca 2ϩ depletion of the ER has little effect on the membrane potential (20) and insulin release (24) * This work was supported by grants from the Swedish Research Council, the from ␤-cells exposed to a sub-stimulatory glucose concentration (3 mM), such depletion further depolarizes glucose-stimulated ␤-cells (23) and amplifies insulin secretion (23,24). Activation of SOCE has much more important effects on the ␣-cell by triggering Ca 2ϩ entry through voltage-operated Ca 2ϩ channels (VOCCs) (22) and glucagon release (24), and SOCE has been attributed to be a central function in epinephrine stimulation and glucose inhibition of glucagon secretion (22,24).
Studies of the store-operated mechanism in primary ␤-cells (21,25) and ␣-cells (22) have been based on measurements of [Ca 2ϩ ] i in Ca 2ϩ omission-readdition (21,22,25) and Mn 2ϩ quench (21,25) experiments. Direct studies of molecules involved in the store-operated mechanism have so far been restricted to clonal MIN6 ␤-cells transfected with STIM1 tagged with enhanced yellow fluorescent protein (STIM1-YFP) showing that Ca 2ϩ depletion of the ER triggers the expected PM association of the molecule (26). We have now utilized adenoviruses encoding STIM1-YFP and Orai1-mCherry (16) to infect pancreatic islets and study STIM1 translocation and association with Orai1 in primary pancreatic islet cells during conditions known to modulate hormone secretion and SOCE. Consistent with a role of SOCE in glucagon secretion, the Ca 2ϩ entry was controlled by similar low glucose concentrations in ␣-cells as those regulating release of the glucose-elevating hormone. We also discovered that cAMP triggers STIM1 translocation to the subplasmalemmal regions but neither induces coclustering of STIM1-Orai1 nor the activation of SOCE that occurs after calcium depletion of the ER. The data indicate that STIM1 translocation can occur independent of Orai1 clustering and SOCE.
Islet Isolation, Cell Culture, and Virus Infection-All procedures for animal handling, preparation, and use of pancreatic islets were approved by local animal and human ethical committees. Islets of Langerhans were isolated from C57BL6J female mice. The animals were placed in a sealed container into which a stream of CO 2 was delivered. When the animals became unconscious, they were exsanguinated by decapitation. After opening the peritoneal cavity, the splenic part of the pancreas was excised and cut into small pieces, which were digested with 1 mg/ml collagenase to obtain free islets of Langerhans. Human pancreatic islets from five normoglycemic cadaveric donors (two female, three male; aged 39 -57 years) were generously provided by the Nordic Network for Clinical Islet Transplantation. The islets were isolated with semiautomated digestion filtration (27) and purified on a continuous density gradient in a refrigerated cell processor (COBE 2191; COBE Blood Component Technology, Lakewood, CO). After purification, the islets were kept for 2-5 days at 37 ºC in an atmosphere of 5% CO 2 in CMRL 1066 culture medium (Mediatech, Herndon, VA) containing 5.5 mM glucose and supplemented with 10 mM nicotinamide, 10 mM HEPES, 0.25 g/ml Fungizone, 50 g/ml gentamicin, 2 mM glutamine, 10 g/ml ciprofloxacin, and 10% fetal calf serum. The islets were subsequently cultured for 1-4 days in RPMI 1640 medium containing 5.5 mM glucose and supplemented with 10% fetal calf serum, 100 g/ml penicillin, and 100 g/ml streptomycin. The islets were then infected with adenovirus encoding STIM1-YFP and/or Orai1-mCherry using a multiplicity of infection of 10 5 fluorescent focus-forming units/islet in culture medium. After 1 h of incubation at 37 ºC, the inoculum was removed, and the islets were washed twice, followed by further culture for 24 h. To check that fluorescence changes did not merely reflect alterations of membrane properties and/or cell adhesion, some measurements were performed after co-infection with adenovirus encoding cyan fluorescent protein (CFP) anchored to the PM as reference (28). Before the experiments, the islets were transferred to a buffer containing 125 mM NaCl, 4.8 mM KCl, 1.3 mM CaCl 2 , 1.2 mM MgCl 2 , and 25 mM HEPES (with pH adjusted to 7.40 with NaOH) and incubated for 30 min at 37 ºC. The islets were then allowed to attach to the center of polylysine-coated round 25-mm coverslips for 5 min. Some experiments were performed on single cells prepared by shaking the freshly isolated islets in a Ca 2ϩ -deficient medium (29). After resuspension in the RPMI 1640 culture medium, the cells were allowed to attach to the center of round coverslips during 2-5 days of culture at 37 ºC in an atmosphere of 5% CO 2 in humidified air.
Measurements of STIM1-YFP Translocation and Redistribution of Orai1-mCherry-The subcellular distribution of the STIM1-YFP fluorescence was analyzed with a Yokogawa CSU-10 spinning disk confocal system (Andor Technology, Belfast, Northern Ireland) attached to a TE2000 microscope (Nikon) with a ϫ60 1.40 NA objective (Nikon). The fluorescence was excited by the 514-nm line of an argon ion laser (ALC ϫ60, Creative Laser Corp, Munich, Germany). The laser beam was homogenized and expanded by a rotating light-shaping diffuser (Physical Optics Corp., Torrance, CA) before being refocused into the confocal scanhead. Fluorescence emission was selected by a 560-/40-nm half-bandwidth filter (Semrock, Rochester, NY) and detected with a back-illuminated EMCCD camera (DU-888, Andor Technology) under MetaFluor (Molecular Devices Corp., Downington, PA) software control. An Eclipse Ti microscope (Nikon) with a total internal reflection fluorescence (TIRF) illuminator and a ϫ60 1.45 NA objective was used for measurements of the PM concentrations of STIM1-YFP, CFP-tagged PM marker, and Orai1-mCherry. The 514-and 458-nm lines of an argon laser (Creative Laser Production) were used for excitation of YFP and CFP, respectively, and the 561-nm line of a diode-pumped solid-state laser (Jive, Cobolt AB, Stockholm, Sweden) was used to excite mCherry. However, because of the properties of the dichroic mirror, the 488-nm line of the argon laser was used to excite YFP when measured in parallel with mCherry. Wavelengths were selected by interference filters (Semrock) mounted in a filter wheel (Sutter Instruments, Novato, CA), and the beam was coupled to the TIRF illuminator by an optical fiber (Oz Optics, Ottawa, Canada). Fluorescence was detected with a back-illuminated EMCCD camera (DU-887, Andor Technology) controlled by the MetaFluor software. Emission wavelengths were selected with interference filters (560-/40-nm half-bandwidth for YFP, 485-/25-nm for CFP; Semrock) or a glass filter (645-nm long pass for mCherry; Melles Griot) mounted in a filter wheel (Sutter Instruments). YFP or mCherry images and YFP/CFP or YFP/mCherry image pairs were acquired every 5 s. The beam was blocked by a shutter (Sutter Instruments) between image captures to minimize exposure of the cells to the potentially harmful laser light. The coverslips with the attached islets were used as exchangeable bottoms of an open custom-built 50-l laminar flow chamber. The cham-ber holder on the microscope stage and the objective were thermostated at 37 ºC. Islets in the chamber were superfused with medium at a rate of 0.3 ml/min.

Measurements of [Ca 2ϩ ] i -For [Ca 2ϩ
] i measurements, the cells were preincubated in the presence of 1 M of the acetoxymethyl ester of Fura-PE3. [Ca 2ϩ ] i imaging was performed with an inverted microscope (Nikon Diaphot) placed in a climate box maintained at 37 ºC. The microscope was equipped for epifluorescence fluorometry with a 400-nm dichroic mirror and a ϫ40 1.3 NA Fluor oil immersion objective. Excitation light was delivered through a 5-mm diameter liquid light guide from an Optoscan monochromator (Cairn Research Ltd., Faversham, UK) with a 150-watt xenon arc lamp. The monochromator provided excitation light at 340 nm (2.5-nm halfbandwidth) and 380 nm (1.9-nm half-bandwidth), and emission was measured at 510 nm (40-nm half-bandwidth) using a EMCCD camera (DU-887, Andor Technology). The Metafluor software controlled the monochromator and the camera, acquiring image pairs every 2 s with 80 -100-ms integration at each wavelength and Ͻ1 ms for changing wavelength and slits. To reduce photodamage, the specimens were illuminated only during image capture. Ratio frames were calculated after background subtraction, and [Ca 2ϩ ] i was estimated as described previously (30).  Cell Identification-Immediately after experiments, ␣and ␤-cells remaining in position within the experimental chamber in the microscope were identified by immunostaining for glucagon and insulin. The cells were fixed by sequential 5-min exposures to 25, 50, 75, and 95% ethanol. After sequential rinsing with 3% H 2 O 2 and Tris buffer (0.05 M, pH 7.4), protein block was added to reduce background staining. After 10 min, polyclonal rabbit anti-glucagon or guinea pig anti-swine insulin (1:100; DAKO) was added for 30 min followed by rinsing with Tris buffer. The MACH 3 TM rabbit probe alkaline phosphatase polymer kit was then used for the visualization according to manufacturer's instructions. After further rinsing with Tris buffer and distilled water, cell nuclei were stained with hematoxylin for 0.5-2 min. The ␣-cells are smaller than the ␤-cells, and the two cell types show opposite responses to epinephrine with regard to STIM1-YFP translocation between the bulk ER and the subplasmalemmal junctions (see "Results"). Therefore, the size of the cell footprint together with the translocation response to epinephrine was used for cell identification in most experiments. These criteria should eliminate the small somatostatin-releasing ␦-cells with ␤-cell-like domination of ␣ 2 -adrenoceptors (31) as small cells and cells with small footprints were never taken as ␤-cells. Because epinephrine mobilizes intracellular Ca 2ϩ in ␣but not ␤-cells (22), this response together with cell size was used for cell identification in most measurements of [Ca 2ϩ ] i .
Data and Statistical Analysis-Image analysis was made using the MetaFluor or ImageJ (W. S. Rasband, National Institutes of Health, rsb.info.nih.gov) software. The STIM1-YFP concentration in the subplasmalemmal region was evaluated as the fluorescence intensity F in relation to the initial fluorescence intensity F 0 after subtraction of background (F/F 0 ). To quantify the redistribution of Orai1 in the PM, we analyzed the intensity variability in the Orai1 images by calculating the variation coefficient of pixel intensities. The pixel size was 266 nm, which is close to the theoretical 222-nm optical resolution in 645 nm of light. Data are presented as means Ϯ S.E. Statistical comparisons were assessed with Student's t test.

Depletion of ER Ca 2ϩ Induces Subplasmalemmal STIM1
Accumulation-Peripheral cells in isolated mouse islets expressing STIM1-YFP and exposed to 3 mM glucose showed diffuse fluorescence over the cytoplasm in confocal microscopy ( Fig. 1A), but also some subplasmalemmal fluorescence puncta were observed with TIRF microscopy (Fig. 1B). Depletion of the ER Ca 2ϩ stores by inhibition of the sarcoendoplasmic reticulum Ca 2ϩ -ATPase (SERCA) with CPA induces gradual formation of much more conspicuous subplasmalemmal puncta (52 Ϯ 4-s rise time; Fig. 1, A and B). This CPA effect was delayed 118 Ϯ 21 s (n ϭ 13). Omission of extracellular Ca 2ϩ with addition of EGTA induced a less marked subplasmalemmal accumulation of STIM1-YFP that was further enhanced by CPA (Fig. 1C).
Epinephrine Induces Opposite Translocation of STIM1 in ␣and ␤-Cells-In the presence of 3 mM glucose, the addition of 5 M epinephrine induced opposite responses in different cells. In the majority of the cells (27 of 41), TIRF microscopy revealed epinephrine-induced loss of PM-associated STIM1-YFP fluorescence indicating retranslocation of the protein from the subplasmalemmal region into the bulk ER ( Fig. 2A). In 5 of the 41 cells that had smaller footprints, epinephrine instead induced pronounced subplasmalemmal accumulation of STIM1-YFP with formation of distinct puncta. Fig. 2B shows that one of the latter cells also responded to CPA with similar STIM1-YFP translocation to the PM. There was no epinephrine response in the remaining nine cells (data not shown). Immunostaining for  insulin and glucagon revealed that the epinephrine-induced decrease and increase of subplasmalemmal STIM1-YFP occurred in ␤-cells (data not shown) and ␣-cells (Fig. 2C), respectively.
Glucose Stimulates Retranslocation of STIM1 to Bulk ER-Increase of the glucose concentration from 3 to 20 mM resulted in loss of subplasmalemmal STIM1-YFP fluorescence in ␤-cells identified by size and epinephrine response (21 of 32; Fig. 3). In most cases, this glucose effect was only partially reversed after reintroduction of 3 mM glucose (Fig. 3A), and rapid restoration of subplasmalemmal STIM1-YFP fluorescence required glucose omission (Fig. 3, B and C). Some ␤-cells (19 of 167) showed oscillations of subplasmalemmal STIM1-YFP fluorescence in 3 mM glucose that were reversibly inhibited by 20 mM of the sugar (all of five ␤-cells; Fig. 3D). Fig. 3E exemplifies the glucose concentration dependence of the loss of subplasmalemmal STIM1-YFP fluorescence, and Fig. 3F summarizes similar observations in nine ␤-cells with half-maximal and maximal effects at 3.4 and 11 mM, respectively.
Also size-and epinephrine-identified ␣-cells responded to glucose with reduction of PM-associated STIM1-YFP fluorescence (Fig. 4A). Because this effect was maximal when glucose had been increased from 0 to 3 mM, we tested additional concentrations within this range. Fig. 4, B and C, shows two ␣-cells cAMP Regulates STIM1 but Neither Orai1 nor SOCE starting to respond to reduction of PM-associated STIM1-YFP fluorescence at 1 and 1.5 mM glucose, respectively, and Fig. 4D summarizes the concentration dependence with half-maximal and maximal effects at 1.3 and 3 mM. Some ␣-cells showed oscillations in subplasmalemmal STIM1-YFP fluorescence in glucose-free medium (6 of 21; Fig. 4C), and these oscillations were inhibited by glucose. The epinephrine response was not always sustained like in the ␣-cells shown in Figs. 2, A and B,  and 4, B and C. In 7 of 51 ␣-cells, the response was instead transient (Figs. 4A, and 6B).
The glucose-induced STIM1-YFP retranslocation from the subplasmalemmal region to the bulk ER was also studied in human pancreatic islets. Fig. 5, A and B, shows traces obtained with epinephrine-identified ␤-cells and that in Fig. 5C from an ␣-cell. Like in mouse ␤-cells, glucose induced concentrationdependent translocation, but at Ն7 mM the gradual reduction of subplasmalemmal fluorescence was interrupted by peaks of PM fluorescence in all five ␤-cells (Fig. 5A). This phenomenon may reflect activation of the store-operated pathway by Ca 2ϩinduced Ca 2ϩ release from the ER triggered by Ca 2ϩ entry through VOCCs. We therefore repeated experiments under conditions preventing such entry. As shown in Fig. 5B, increasing glucose concentrations monotonously reduced subplasmalemmal STIM1-YFP fluorescence in the presence of the hyperpolarizing ATP-sensitive K ϩ (K ATP ) channel activator diazoxide with maximal effect at 7-11 mM of the sugar in two studied ␤-cells. Also, four human ␣-cells behaved similarly to mouse ␣-cells with maximal loss of subplasmalemmal fluorescence already with 3 mM glucose (Fig. 5C).
cAMP Induces Subplasmalemmal STIM1 Accumulation Involving PKA and Epac-Because epinephrine acts on ␣ 1 -(32) and ␤-adrenoceptors (32,33) in ␣-cells to increase IP 3 and cAMP (34) resulting in Ca 2ϩ release from the ER, it was not surprising that epinephrine induced subplasmalemmal accumulation of STIM1-YFP. However, the opposite effect in ␤-cells was unexpected considering that epinephrine acts on ␣ 2 -adrenoceptors (33,35) to lower cAMP (34). We therefore tested whether changes in cAMP might be involved in the STIM1 translocation responses of the two cell types. These experiments were done at a basal glucose concentration to prevent cAMP from promoting IP 3 receptor-mediated Ca 2ϩ release, which only occurs in ␤-cells exposed to higher concen-  trations of the sugar (36,37). Because [Ca 2ϩ ] i in ␤-cells is low and stable under these conditions, the reported effects of [Ca 2ϩ ] i elevation on STIM1 translocation (38) should not influence the results. Fig. 6, A and B, shows that the rise of cAMP by activation of adenylyl cyclases with 5 M forskolin increased STIM1-YFP translocation to the PM in 22 of 27 ␤-cells with opposite response to epinephrine. Also, most epinephrineidentified ␣-cells (6 of 8) reacted to forskolin in a similar manner (Fig. 6B), and the remaining ␣-cells did not respond (Fig.  6A). Fig. 6, C and D, illustrates that STIM1-YFP translocation to the PM after rise of cAMP by phosphodiesterase inhibition with 50 M IBMX is reversed by 100 M of the adenylyl cyclase inhibitor DDA in all of 10 ␤-cells (Fig. 6C) and all of 9 ␣-cells (Fig. 6D) identified with epinephrine. Also, the specific PKA agonists N 6 -phenyladenosine-3Ј,5Ј-cyclic monophosphate (100 M) induced STIM1-YFP translocation to the PM in 11 of 17 ␤-cells (Fig. 6E) and 6 of 10 ␣-cells (data not shown) identified with epinephrine, and similar effects were seen with another PKA activator (100 M Sp-8-CPT-cAMPS; data not shown). However, elevation of cAMP with forskolin induced additional PM translocation of STIM1-YFP (Fig. 6E, 6

of 8 ␤-cells).
We therefore also tested the effect of 1-2 M of the Epac activator 8-(4chlorophenylthio)-2Ј-O-methyladenosine-3Ј,5Ј-cyclic monophosphate, acetoxymethyl ester, which also induced STIM1-YFP translocation to the PM in 16 of 26 ␤-cells (Fig. 6F) and 7 of 11 ␣-cells (data not shown) identified with epinephrine. Our data therefore indicate that both PKA and Epac contribute to mediate the effect of cAMP on STIM1-YFP translocation to the PM.
cAMP-stimulated STIM1 Translocation Does Not Activate SOCE-The effects of cAMP modulation on SOCE were studied using a Ca 2ϩ omission-readdition approach in cells that were hyperpolarized with diazoxide and exposed to the Ca 2ϩ channel blocker methoxyverapamil to prevent voltage-operated Ca 2ϩ entry. The ␤-cell in Fig. 7A was initially exposed to 1.3 mM Ca 2ϩ and 3 mM glucose, which causes less than halfmaximal Ca 2ϩ filling of the ER (19) associated with partial inactivation of the store-operated pathway (21,25). When subsequent omission of extracellular Ca 2ϩ had lowered [Ca 2ϩ ] i to a stabile level, the introduction of 10 mM Ca 2ϩ induced 18 Ϯ 4 nM cAMP Regulates STIM1 but Neither Orai1 nor SOCE (n ϭ 7) elevation of [Ca 2ϩ ] i . The effect of such Ca 2ϩ omissionreaddition was not altered by forskolin (17 Ϯ 4 nM increase of [Ca 2ϩ ] i ; n ϭ 7). When the SOCE was fully activated by emptying the ER with CPA, Ca 2ϩ omission-readdition raised [Ca 2ϩ ] i by 66 Ϯ 8 nM (n ϭ 7; p Ͻ 0.005). Also under these conditions forskolin failed to affect the response (67 Ϯ 9 nM; n ϭ 7).
The effect of cAMP elevation with forskolin on SOCE was also tested in the presence of 20 mM glucose that maximally fills the ER with Ca 2ϩ (19) and inhibits the store-operated pathway (21,25) as well as under conditions when this pathway was activated with CPA (Fig. 7B). Again, forskolin had no effect on the Ca 2ϩ readdition-induced elevation of [Ca 2ϩ ] i under either of these conditions. Before addition of CPA, the increase was 15 Ϯ 5 nM both in the absence and presence of forskolin, and after addition of CPA the increases were 55 Ϯ 4 and 50 Ϯ 4 nM (n ϭ 6) in the absence and presence of forskolin, respectively. cAMP-induced accumulation of subplasmalemmal STIM1 apparently occur without effect on SOCE.
cAMP-stimulated STIM1 Translocation Does Not Induce STIM1-Orai1 Co-clustering-Searching for an explanation why cAMP-induced STIM1 translocation did not activate SOCE, we next studied Orai1 distribution in the PM of islet cells expressing Orai1-mCherry alone (data not shown) or in combination with STIM1-YFP (Fig. 8). When the SERCA pump was ener-gized by the presence of 20 mM glucose to fill the ER with calcium and the cells hyperpolarized with diazoxide to prevent triggering of Ca 2ϩ -induced Ca 2ϩ release, there were few STIM1-YFP puncta at the PM (Fig. 8A, top panels). The number of puncta increased markedly in response to forskolin, and the effect was even more striking after depletion of ER Ca 2ϩ with CPA. The more pronounced effect of CPA than of forskolin was preferentially observed when STIM1 was co-expressed with Orai-1. In cells expressing STIM1-YFP alone, the total PM-associated STIM1-YFP fluorescence increased 1.64 Ϯ 0.19-fold in the presence of forskolin and 1.82 Ϯ 0.21-fold (not significant, n ϭ 6) of control during subsequent exposure to CPA. However, after co-expression with Orai1-mCherry, the total PM-associated STIM1-YFP fluorescence in response to forskolin was 1.31 Ϯ 0.16-fold and that induced by CPA was 2.25 Ϯ 0.24-fold (p Ͻ 0.001; n ϭ 8) of control.
Under control conditions, Orai1-mCherry showed a diffuse distribution with even PM fluorescence and no apparent effect of exposure to forskolin. The variation coefficients in pixel intensities under these conditions were identical (10.2 Ϯ 0.3% for control and 10.3 Ϯ 0.2% for forskolin). However, subsequent SERCA inhibition with CPA induced Orai1-mCherry redistribution with formation of a punctate pattern (Fig. 8A, middle panels) resulting in a highly significant (p ϭ 0.001) increase of the variation coefficient to 13.5 Ϯ 0.4%. The CPAinduced Orai1-mCherry puncta co-localized with those of STIM1-YFP, whereas STIM1-YFP puncta formed in response to forskolin did not associate with increased Orai1-mCherry (Fig. 8A, bottom panels). To quantify the changes in STIM1 and Orai1, regions of interest (ROIs) were selected corresponding to STIM1 puncta formed either in response to CPA or forskolin, and the STIM1-YFP and Orai1-mCherry fluorescence was then measured in these ROIs. Fig. 8B shows that the STIM1-YFP fluorescence within ROIs defined by CPA-induced STIM1-YFP puncta showed a less pronounced increase during exposure to forskolin. However, the Orai1-mCherry fluorescence within these ROIs only increased in response to CPA (Fig.  8C). When the ROIs were instead defined by STIM1-YFP puncta formed in response to forskolin, the STIM1-YFP fluorescence within these ROIs also increased after exposure to CPA (Fig. 8D). The Orai1-mCherry fluorescence within the same ROIs was not different under basal conditions and in the presence of forskolin but tended to decrease slightly after exposure to CPA (Fig. 8E). Orai1 consequently only associates with STIM1 after Ca 2ϩ depletion of the ER and preferentially at sites other than those where STIM1 forms puncta in response to forskolin. This conclusion was supported by comparing the location of STIM1 puncta formed in response to forskolin and CPA. As seen in Fig. 8F, there was relatively little overlap (yellow) between STIM1-YFP puncta formed in response to forskolin (displayed in red) and those formed after exposure to CPA (displayed in green).

DISCUSSION
This study shows that Ca 2ϩ release from the ER and sequestration of the ion in this organelle in the insulin-and glucagonreleasing pancreatic islet cells cause the characteristic translocations of STIM1 to and from the subplasmalemmal junctions, FIGURE 7. Effects of cAMP elevation on SOCE. Epifluorescence recordings of [Ca 2ϩ ] i in Fura-PE3-loaded large islet cells exposed to 250 M of the K ATP channel activator diazoxide. Apart from the size, the cells were identified as ␤-cells based on the lack of response to epinephrine (Epi, 5 M). A, forskolin does not affect SOCE in a cell exposed to 3 mM glucose. VOCCs were blocked by hyperpolarization with diazoxide and presence of 100 M methoxyverapamil. The [Ca 2ϩ ] i response to Ca 2ϩ omission followed by exposure to 10 mM Ca 2ϩ was studied under control conditions and in the presence of 5 M forskolin before and after full activation of SOCE with 100 M CPA. B, forskolin does not affect SOCE in a cell exposed to 20 mM glucose. Like in A, hyperpolarizing diazoxide and the voltage-operated Ca 2ϩ channel blocker methoxyverapamil were present throughout. The [Ca 2ϩ ] i response to Ca 2ϩ omission followed by exposure to 10 mM Ca 2ϩ was studied under control conditions and in the presence of 5 M forskolin before and after full activation of SOCE with 100 M CPA.

cAMP Regulates STIM1 but Neither Orai1 nor SOCE
respectively. It also demonstrates that epinephrine mimics the effect of Ca 2ϩ store depletion in inducing subplasmalemmal accumulation of STIM1 in ␣-cells but that the opposite effect is observed in ␤-cells. Searching for the underlying mechanism, we found that cAMP, which increases and decreases in response to epinephrine in ␣and ␤-cells, respectively (34), is involved in the subplasmalemmal accumulation of STIM1. Because the dynamin-related mitochondrial protein mitofusin 2, which was recently implicated in the trafficking of STIM1 to the ER-PM junctions (39), is regulated by PKA (40), we speculate that this mechanism partakes in the cAMP-induced STIM1 translocation. Additional processes are likely involved, as translocation was induced both by specific PKA and Epac agonists. However, although the STIM1 translocation determined by the filling state of the ER affected STIM1-Orai1 co-clustering and modulated SOCE as expected, translocation determined by cAMP occurred without effect on Orai1 and Ca 2ϩ entry. Activation of SOCE by Ca 2ϩ store depletion has been found to involve a conformational transition that releases the Orai-activating region of STIM1 from an intramolecular clamp (41,42). Our data indicate that cAMP induces STIM1 translocation independent of such a conformational change.
Glucose is the major physiological stimulator of insulin secretion. The sugar is taken up and metabolized by the ␤-cells resulting in increased ATP production. An early effect of glucose stimulation is therefore to energize the SERCA pump causing calcium sequestration in the ER, lowering of [Ca 2ϩ ] i (20,43) and initial inhibition of insulin secretion (44,45). The increased ATP/ADP ratio also closes K ATP channels in the PM with ensuing depolarization. Subsequent opening of L-type VOCCs results in entry of Ca 2ϩ and a rise of [Ca 2ϩ ] i that trig-FIGURE 8. Ca 2؉ store depletion but not cAMP induces STIM1-Orai1 co-clustering in islet cells. A, TIRF images of an islet cell co-expressing STIM1-YFP (green) and Orai1-mCherry (red). The cell was initially exposed to 20 mM glucose while hyperpolarized with 250 M of the K ATP channel activator diazoxide (basal). Subsequently 5 M forskolin and 50 M CPA were added. The top and middle rows show STIM1 and Orai1 images, respectively, and the bottom row shows a STIM1/Orai1 overlay. The images represent averages of 25 consecutive frames under each condition. B and C, changes in mean STIM1-YFP and Orai1-mCherry fluorescence intensity Ϯ S.E. in 25 ROIs based on selection of STIM1 puncta that were formed after exposure to CPA. D and E, changes in mean STIM1-YFP and Orai1-mCherry fluorescence intensity Ϯ S.E. in 18 ROIs based on selection of STIM1 puncta that were formed after exposure to forskolin. F, overlay image of STIM1 after exposure to forskolin (red) and CPA (green). Image scale bars indicate 1 m. gers a pronounced peak of secretion replacing the initial inhibition (46). ER sequestration of Ca 2ϩ is half-maximal and maximal at about 6 and 20 mM glucose, respectively (19,(47)(48)(49). It is well documented that Ca 2ϩ emptying of the ER activates SOCE in ␤-cells (19 -21, 23, 25, 50), and the rate of SOCE is inversely dependent on the Ca 2ϩ filling of the ER in a graded fashion (25). The presently observed glucose-induced STIM1 retranslocation from the PM to the cell interior in mouse ␤-cells showed somewhat higher glucose sensitivity than Ca 2ϩ filling of the ER with half-maximal and maximal effects at 3.4 and 11 mM, respectively. Human ␤-cells were similar in this respect, but Ca 2ϩ -induced Ca 2ϩ release may be more prominent in the human cells explaining why peaks of near-membrane STIM1-YFP fluorescence interrupted glucose-induced STIM1 disappearance from the subplasmalemmal region. The physiological relevance of the glucose effects on ER Ca 2ϩ filling in ␤-cells is probably to secure a pool of releasable Ca 2ϩ rather than regulating the store-operated pathway, which has modest effects on [Ca 2ϩ ] i and insulin secretion (20,21,23,25).
The glucose sensitivity of STIM1 retranslocation to the ER was strikingly higher in ␣than in ␤-cells with maximal effect at 3 mM reinforcing indirect observations that this concentration is sufficient for maximal Ca 2ϩ filling of the ␣-cell ER (22). The different glucose sensitivities cannot be attributed to the fact that ␣-cells only express SERCA2b, whereas the ␤-cell also express the lower affinity Ca 2ϩ transporter SERCA3 (51). The glucose concentration dependence of ER Ca 2ϩ filling in ␤-cells is thus unaffected after SERCA3 knock-out (52). SOCE in ␣-cells is consequently controlled by the same low glucose concentrations that regulate glucagon release. Because ␣-cells have a higher input resistance than ␤-cells, the membrane potential is sensitive to small changes in current (53). Therefore, SOCE has a much more pronounced effect in ␣-cells, depolarizing the membrane sufficiently to trigger Ca 2ϩ influx through VOCCs (22) and glucagon release (24). Epinephrine stimulates glucagon secretion in this manner by binding to ␣ 1 -and ␤-adrenoceptors (32) leading to ER release of Ca 2ϩ with resulting SOCE and VOCC activation (22). We have proposed that glucose also modulates glucagon secretion via the store-operated mechanism. Under normoglycemic conditions the ER is filled with Ca 2ϩ , and there is no SOCE activation and thus no depolarization to activate the VOCCs. By contrast, when glucose declines to hypoglycemic levels, the ER begins to empty, and the resulting SOCE activation provides the membrane depolarization that triggers VOCC activation and glucagon release (22,24).
In summary, this study shows that STIM1 accumulates in the subplasmalemmal region in response to Ca 2ϩ depletion of the ER in ␣and ␤-cells and that glucose-induced Ca 2ϩ refilling of the ER stimulates the opposite translocation. The observation that a glucose concentration as low as 3 mM causes maximal retranslocation of STIM1 to the ER in ␣-cells is consistent with the idea that the sugar inhibits glucagon secretion by shutting off the store-operated pathway. We also found that cAMP stimulates subplasmalemmal accumulation of STIM1 in both ␣and ␤-cells. However, the cAMP-induced STIM1 puncta formed without co-clustering of Orai1 or activation of SOCE. Apparently, factors in addition to STIM1 translocation are required for the formation of functional Orai1 channels and activation of SOCE.