Ca 2 (cid:1) -induced Ca 2 (cid:1) Release via Inositol 1,4,5-trisphosphate Receptors Is Amplified by Protein Kinase A and Triggers Exocytosis in Pancreatic (cid:2) -Cells*

Hormones, such as glucagon and glucagon-like pep-tide-1, potently amplify nutrient stimulated insulin secretion by raising cAMP. We have studied how cAMP affects Ca 2 (cid:1) -induced Ca 2 (cid:1) release (CICR) in pancreatic (cid:2) -cells from mice and rats and the role of CICR in secretion. CICR was observed as pronounced Ca 2 (cid:1) spikes on top of glucose- or depolarization-dependent rise of the cytoplasmic Ca 2 (cid:1) concentration ([Ca 2 (cid:1) ] i ). cAMP-elevat-ing agents strongly promoted CICR. This effect involved sensitization of the receptors underlying CICR, because many cells exhibited the characteristic Ca 2 (cid:1) spiking at low or even in the absence of depolarization-dependent elevation of [Ca 2 (cid:1) ] i . The cAMP effect was mimicked by a specific activator of protein kinase A in cells unrespon-sive to activators of cAMP-regulated guanine nucleotide exchange factor. Ryanodine pretreatment, which abolishes CICR mediated by ryanodine receptors, did not prevent CICR. Moreover, a high concentration of caffeine, known to activate ryanodine receptors independ-ently of Ca 2 (cid:1) , failed to mobilize intracellular Ca 2 (cid:1) a concentration CICR 1,4,5-trisphosphate receptors 3 Rs). the IP borate blocked the cAMP-promoted CICR. Individual CICR events in pancreatic (cid:2) -cells were followed by [Ca 2 (cid:1) ] i spikes in neighboring human erythroleukemia cells, used to report secretory events in the (cid:2) -cells. The results indicate that protein kinase A-mediated promotion of CICR via IP 3 Rs is part of the mechanism by which cAMP amplifies insulin release.

the insulin-containing granules (1). By stimulating Ca 2ϩ sequestration in the endoplasmic reticulum (ER) (2)(3)(4), glucose also has an important role in preparing the ␤-cell to respond to hormones and neurotransmitters, which act by mobilizing Ca 2ϩ from the ER (5)(6)(7). The latter effects are in most cases caused by activation of phospholipase C, catalyzing the formation of inositol 1,4,5-trisphosphate (IP 3 ). The IP 3 receptor (IP 3 R) is a Ca 2ϩ channel in the ER membrane (8). Another putative pathway for Ca 2ϩ release from the ER is via ryanodine receptors (RyRs). Although RyRs are expressed in ␤-cells (9 -12), their physiological role remains controversial (12)(13)(14). Ca 2ϩ -induced Ca 2ϩ release (CICR) is a mechanism by which any local rise of [Ca 2ϩ ] i becomes further amplified by Ca 2ϩ release from stores. The heart is a classic example of CICR, where it provides a link between depolarization-dependent influx of "triggering" Ca 2ϩ and release of contraction-inducing Ca 2ϩ from the sarcoplasmic reticulum (15). In heart cells, CICR is caused by activation of RyRs. However, in many other types of cells, IP 3 Rs are equally competent in mediating CICR because they display a similar autocatalytic Ca 2ϩ release mechanism (16). The binding of IP 3 thus sensitizes the IP 3 Rs to the stimulatory effect of Ca 2ϩ (17,18).
As in the heart, CICR in the ␤-cell may provide a link between influx of Ca 2ϩ and release from intracellular stores, resulting in amplification of the Ca 2ϩ signal triggering insulin secretion (19). Several studies propose that CICR in ␤-cells is mediated by RyRs (10, 19 -22). Critical experiments in the latter studies rely on the use of tumor-transformed clonal ␤-cells, and we have recently confirmed the expression of functional RyRs in rat insulinoma cells (23). However, our study also showed that CICR in primary ␤-cells from mice, rats, and human subjects is caused by activation of IP 3 Rs rather than RyRs. Agents raising cAMP have been found to promote intracellular Ca 2ϩ mobilization in insulin-releasing cell lines and pancreatic ␤-cells, and this action was proposed to be mediated by sensitization of either IP 3 Rs (24 -26) or RyRs (9, 10, 20 -22, 26, 27). Phosphorylation of the RyRs by the cAMP-dependent protein kinase A (PKA) was assumed to be a prerequisite for CICR (9,10). However, in later studies, a PKA-independent pathway involving cAMP-regulated guanine nucleotide ex-* This work was supported by Grant 06240 from the Swedish Research Council, and grants from the Swedish Diabetes Association, the Scandinavian Physiological Society, the Family Ernfors foundation and the Knut and Alice Wallenberg Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed. change factor (Epac) has been suggested (20,22,27). Because important evidence for this concept was obtained with clonal ␤-cells, we have now studied the mechanisms by which cAMP promotes CICR in primary mouse and rat ␤-cells. We show that cAMP-facilitated CICR is caused by PKA-dependent activation of IP 3 Rs. Moreover, our data indicate that CICR is part of the mechanism by which cAMP amplifies insulin release.
Preparation and Culture of Cells-Islets of Langerhans were collagenase-isolated from pieces of pancreas from ob/ob mice or Wistar rats. Free cells were prepared by shaking the islets in a Ca 2ϩ -deficient medium (28). The cells were suspended in RPMI 1640 medium supplemented with 10% fetal calf serum, 100 IU/ml penicillin, 100 g/ml streptomycin and 30 g/ml gentamicin and allowed to attach to circular 25-mm cover slips during 1-3 days in culture at 37°C in a humidified atmosphere of 5% CO 2 . The ob/ob mouse islets contain more than 90% ␤-cells (29), which respond normally to glucose and other regulators of insulin release (30). The selection of ␤-cells for analysis was based on their large size and low nuclear/cytoplasmic ratio compared with the cells secreting glucagon, somatostatin (31,32), and pancreatic polypeptide (33). Human erythroleukemia 92.1.7 (HEL) cells were obtained from Prof. K. E. O. Åkerman (Uppsala, Sweden) and cultured in suspension in RPMI 1640 medium (34).
Image Analysis of Cytoplasmic Ca 2ϩ -In most experiments, loading of cells with the indicator fura-2 was performed during 30-min incubation at 37°C in a HEPES-buffered medium (25 mM; pH 7.4) containing 0.5 mg/ml bovine serum albumin, 138 mM NaCl  (35). The chamber profile was defined by a 4-mm wide, 7-mm long oval hole in a 1-mm thick silicon rubber gasket with a 25-mm outer diameter. A thin 25-mm diameter stainless steel plate with an identical central opening pressed the rubber gasket to the coverslip by the threaded Sykes-More chamber mount. Inlet and outlet cannulas fixed to the stainless steel plate allowed laminar flow superfusion. The chamber was placed on the stage of an inverted microscope (Eclipse TE2000U; Nikon, Kanagawa, Japan). The chamber holder and the CFI S Fluor 40 ϫ 1.3 numerical aperture oil immersion objective (Nikon) were maintained at 37°C by custombuilt thermostats. The chamber was superfused at a rate of 0.3 ml/min with the loading medium lacking indicator.
The microscope was equipped with an epifluorescence illuminator (Cairn Research Ltd, Faversham, UK) connected through a 5-mm diameter liquid light guide to an Optoscan monochromator (Cairn Research Ltd) with rapid grating and slit width adjustment and a 150watt xenon arc lamp. The monochromator provided excitation light at 340 nm (1.7 nm half-bandwidth) and 380 nm (1.4 nm half-bandwidth). Emission was measured at 510 nm (40 nm half-bandwidth) using a 400-nm dichroic beam splitter and a cooled OrcaER-1394 Firewire digital charge-coupled device camera (Hamamatsu Photonics, Hamamatsu City, Japan) equipped with a C8600 -2 image intensifier (Hamamatsu Photonics). The Metafluor software (Universal Imaging Corp. Downingtown, PA) controlled the monochromator and the chargecoupled device camera, acquiring pairs of images at 340 and 380 nm every 2 s with integration for 60 -80 ms at each wavelength and Ͻ1 ms for changing wavelength and slits. To minimize bleaching and photo damage, the monochromator slits were closed until the start of the next acquisition cycle. Ratio (R) images were calculated after subtraction of background images. [Ca 2ϩ ] i values were obtained according to Grynkiewicz et al. (36) 2ϩ is 224 nM. F 0 and R min are the fura-2 fluorescence at 380 nm and the 340/380 nm fluorescence excitation ratio, respectively, in an "intracellular" K ϩ -rich medium lacking Ca 2ϩ . F S and R max are the corresponding data obtained with a saturating concentration of Ca 2ϩ .
Detection of ␤-cell Secretion-To record secretory events from the mouse ␤-cells, we used HEL cells as reporter cells. The suspensioncultured HEL cells were spun down and loaded with 0.5 M fura-2 as described previously (34). ␤-Cells attached to cover slips were loaded with the single wavelength Ca 2ϩ indicator fluo-4 by incubation in medium (similar to that used for ␤-cells above) containing 11 mM glucose, 250 M diazoxide, and 1 M fluo-4/AM but lacking methoxyverapamil. After loading for 30 min, the HEL-cells and ␤-cell were rinsed in indicator-free medium. The coverslip with ␤-cells was used as bottom of the chamber described above. In this case, a 0.25-mm thick, 25-mm diameter polyester spacer with a central 4-mm wide and 7-mm oval hole (like that in the silicon rubber gasket; see above) was placed on top of the coverslip. Fura-2-loaded HEL cells were pipetted into the central cavity defined by the hole in the spacer, which was then covered by a 25-mm polycarbonate membrane filter before aligning the hole in the silicon rubber gasket and the stainless steel plate (see above) with the hole in the polyester spacer. ␤-Cells attached to the coverslip and free-floating HEL cells under the permeable polycarbonate filter could then be superfused with minimal cell movements. The principal arrangement has been described previously (34), but our chamber had a different geometry. Fluorescence was measured essentially as described above with emission at Ͼ510 nm (long-pass filter) using a 495 nm dichroic beam splitter. Fura-2 fluorescence from the HEL cells was excited at 340 and 380 nm and fluo-4 fluorescence from the ␤-cells at 470 nm. Images were obtained every 2 s with integration for 100 ms at each wavelength and Ͻ1 ms for changing wavelength and slits. Because of the stronger fluorescence from fluo-4, the image intensifier gain was reduced during excitation at 470 nm to balance the signals. [Ca 2ϩ ] i in the HEL cells was calculated as described above. The fluo-4 fluorescence intensity indicating [Ca 2ϩ ] i variations in the ␤-cells was expressed as the ratio between deviation from the basal fluorescence and the basal fluorescence (⌬F/F 0 ).
Statistical Analysis-Only recordings from isolated individual ␤-cells were included in the analyses. Statistical evaluations of the proportion of cells with a certain response were made with Fishers exact test or 2 test with Yates' correction using SigmaStat software (SPSS Inc., Chicago, IL). Wilcoxon signed rank test was used to compare the frequency of Ca 2ϩ spikes. Statistical significance was set at a p value of Ͻ 0.05.

CICR Is Promoted by Glucagon, GLP-1, and other cAMP
Agonists-In accordance with earlier data (37,38), increase of the glucose concentration from 3 to 20 mM induced initial lowering of [Ca 2ϩ ] i in ␤-cells followed by slow, large amplitude oscillations of [Ca 2ϩ ] i (Fig. 1A). The use of low indicator concentrations facilitated the detection of rapid [Ca 2ϩ ] i spikes, which were superimposed on top of the large amplitude oscillations. Addition of diazoxide, which hyperpolarizes the ␤-cells by opening K ATP channels, immediately abolished the slow [Ca 2ϩ ] i oscillations as well as the spikes. In the presence of diazoxide, the membrane potential is near the equilibrium potential for K ϩ . Under these conditions, depolarization with 90 mM KCl induced a rapid rise of [Ca 2ϩ ] i with superimposed [Ca 2ϩ ] i spikes, indicating that the depolarization-dependent influx of Ca 2ϩ triggers CICR. Most subsequent experiments were performed in the presence of 20 mM glucose to stimulate Ca 2ϩ sequestration in the ER (2-4) and diazoxide to keep the membrane potential close to the equilibrium potential for K ϩ .  After return to the physiological K ϩ concentration addition of a low concentration of the adenylyl cyclase activator forskolin evoked occasional CICR spikes (not shown). However, combining 17 mM KCl with forskolin increased the frequency of CICR spiking 4-fold. The ability of cAMP to uncover CICR in response to depolarization-dependent rise of [Ca 2ϩ ] i was not restricted to mouse ␤-cells. Fig. 1, C and D, illustrates [Ca 2ϩ ] i spiking in rat ␤-cells depolarized with 30 mM KCl in the presence of the adenylyl cyclase-activating hormones GLP-1 and glucagon, respectively. However, the mouse ␤-cells were more sensitive, and the same concentrations of glucagon and GLP-1 caused repetitive CICR spiking even in the absence of depolarizationdependent elevation of [Ca 2ϩ ] i . This is illustrated in Fig. 2, which like most subsequent experiments was performed in the presence of methoxyverapamil to block depolarization-dependent Ca 2ϩ entry and keep [Ca 2ϩ ] i at basal levels. Apart from the physiological activation of adenylyl cyclase with glucagon ( Fig.  2A) and GLP-1 (Fig. 2B) CICR spiking from the baseline was also obtained after direct activation of adenylyl cyclase with forskolin (Fig. 2C), by inhibition of cAMP degradation with the phosphodiesterase inhibitors IBMX (Fig. 2D) or caffeine (Fig.  2E), and by cell membrane-permeable 8-Br-cAMP (Fig. 2F).
PKA Rather Than Epac Mediates the cAMP Effect on CICR-Occasional Ca 2ϩ spikes were observed in 2 of 64 individual islet cells exposed to the Epac-specific activator 8-pCPT-2Ј-O-Me-cAMP and in 1 of 57 cells exposed to the even more potent 8-pMeOPT-2Ј-O-Me-cAMP (data not shown). Epac activatorinduced Ca 2ϩ spikes were sometimes seen also in islet cell clusters. These data indicate that the response represents a different cell type than the dominating ␤-cells. However, 111 of 174 cells (64%; p Ͻ 0.001) reacted to the PKA-specific activator Sp-5,6-DCl-cBIMPS with generation of repetitive [Ca 2ϩ ] i spikes. Fig. 3, A and B, illustrate lack of response to the Epac activators 8-pCPT-2Ј-O-Me-cAMP and 8-pMeOPT-2Ј-O-Me-cAMP in cells subsequently reacting to the PKA activator. Further evidence for a PKA mechanism was obtained from the observation that the frequencies of the Ca 2ϩ spiking induced by glucagon and forskolin were reduced by about 70% by the competitive PKA antagonists Rp-8-Br-cAMPS (Fig. 3C) and Rp-8-CPT-cAMPS (Fig. 3D).
Because the caffeine data indicate that cAMP-promoted CICR involves the IP 3 signaling pathway, we tested the effect of the membrane-permeable IP 3 R inhibitor 2-APB (43). The inhibitory effect of 2-APB is incomplete (44), and we found that 50 M prevented Ca 2ϩ signaling induced by 10 M carbachol in 35% of the cells but never the response to 100 M carbachol (Fig. 5A). However, 50 M 2-APB inhibited CICR spiking promoted by glucagon (Fig. 5B) and forskolin (Fig. 5C).
We also investigated whether pretreatment with a high concentration of ryanodine, which abolishes RyR-mediated CICR in clonal ␤-cells (12,23), affects CICR in primary mouse ␤-cells. Ryanodine neither affected CICR in response to depolarizationdependent elevation of [Ca 2ϩ ] i during exposure to 90 mM KCl (Fig. 6A) nor that induced from basal [Ca 2ϩ ] i levels by glucagon (Fig. 6B) or the PKA-specific activator Sp-5,6-DCl-cBIMPS (Fig. 6C). Fig. 6C also shows that addition of 20 mM caffeine inhibits the CICR in response to Sp-5,6-DCl-cBIMPS. The in-hibitory effects of caffeine and 2-APB and the lack of effects of ryanodine indicate that the cAMP-promoted CICR is the result of activation of IP 3 Rs rather than RyRs.
CICR Spikes Trigger Secretion in ␤-cells-To record secretory events from the mouse ␤-cells, we used HEL cells as reporter cells. The HEL cells are not electrically excitable but show robust [Ca 2ϩ ] i responses to many neurotransmitters (34), including ATP (45), which is released together with insulin from the ␤-cell secretory granules (46). Fig. 7 shows recordings of [Ca 2ϩ ] i in two ␤-cells and surrounding HEL cells. The cells were stimulated with 11 mM glucose, and [Ca 2ϩ ] i was kept low in the beginning of the experiment by the presence of diazoxide. Depolarization with 90 mM KCl resulted in elevation of [Ca 2ϩ ] i in the ␤-cells with superimposed spikes caused by CICR. The adjacent HEL cells did not respond to the depolarization, but some HEL cells reacted with [Ca 2ϩ ] i spikes following spikes in the closest ␤-cell. In the ␤-cell labeled ␤-1, the first [Ca 2ϩ ] i spike was followed within one acquisition cycle (2 s) by spikes in the cells labeled HEL-1 and HEL-2 near ␤-1, but these HEL cells did not respond to two subsequent spikes in ␤-1. In the ␤-cell labeled ␤-2, the second and the last [Ca 2ϩ ] i spikes were followed by spikes within 2 s in the HEL-3 cell, which is closest. Moreover, the HEL-4 cells located about one cell diameter away responded to the last spike in the ␤-2 within 4 s. Of 16 Ca 2ϩ spikes in eight HEL cells, they always occurred shortly after spikes in five nearby ␤-cells.

DISCUSSION
The hormones GLP-1 and glucagon potently amplify nutrient-stimulated insulin secretion by raising cAMP, which interacts with a plethora of signal transduction processes, including ion channel activity, intracellular Ca 2ϩ handling, and exocytosis of the insulin-containing granules (47). As shown here and demonstrated elsewhere (24,48), elevation of cAMP promotes [Ca 2ϩ ] i spiking superimposed on depolarization-dependent Ca 2ϩ entry in glucose-stimulated ␤-cells, an effect originally attributed to mobilization of intracellular Ca 2ϩ after sensitization of IP 3 Rs (24) by a PKA mechanism (25). However, when this action of cAMP was first conceptually associated with CICR, it was instead assumed to represent PKA-dependent phosphorylation of RyRs (9,10). Maintaining the idea that RYRs are involved, the role of cAMP was later reconsidered, claiming that the receptor activation is caused by a PKAindependent Epac mechanism (20,22,27).
Although opinions differ with regard to the type of receptors, there seems to be general agreement that cAMP promotes CICR by receptor sensitization. Such a mechanism implies that cAMP enables CICR to be triggered at lower concentrations of Ca 2ϩ . The present data indicate that cAMP not only promotes CICR in response to depolarization-dependent elevation of [Ca 2ϩ ] i but also sensitizes the underlying mechanisms sufficiently for CICR to occur from basal levels of [Ca 2ϩ ] i in ␤-cells exposed to hyperpolarizing diazoxide and to the voltage-dependent Ca 2ϩ channel blocker methoxyverapamil. This is a favorable situation, because CICR can be studied under conditions allowing discrimination between depolarization-dependent influx and intracellular release of Ca 2ϩ , which was not possible with some previously used protocols.
Eliminating interference from depolarization-dependent Ca 2ϩ influx, we first studied whether cAMP acts via PKA or Epac. Our data unequivocally favored PKA. Only 2-3% of the islet cells reacted with one or two Ca 2ϩ spikes when exposed to two potent Epac activators, suggesting that this response originated from cells other than the dominating ␤-cells. However, a specific activator of PKA induced repetitive CICR spikes in 64% of the islet cells. Moreover, two competitive PKA antagonists inhibited CICR spiking in response to cAMP elevation by about 70%. Although a stimulatory effect of CO on Ca 2ϩ spiking in ␤-cells exposed to glucagon has been taken to indicate that cGMP promotes intracellular mobilization of Ca 2ϩ (49), we found no effect of protein kinase G agonists. Protein kinase G activation alone is consequently insufficient for promoting CICR under the present conditions.
We proceeded to study the type of receptor involved in CICR after elevation of cAMP. Caffeine was one tool in this exploration. The classic effect of caffeine on ␤-cells is phosphodiesterase inhibition with elevation of cAMP (50). Therefore, it is not surprising that a low concentration of caffeine mimicked the effect of other cAMP agonists in promoting CICR. However, caffeine also sensitizes RyRs to Ca 2ϩ (42), and high concentrations even activate RyRs independent of Ca 2ϩ (41). Our observation that 20 mM caffeine failed to mobilize intracellular Ca 2ϩ in ␤-cells exposed to cAMP agonists argues against the involvement of RyRs in CICR. This conclusion was further supported by the lack of effect of ryanodine pretreatment on CICR induced by depolarization-dependent elevation of [Ca 2ϩ ] i , by elevation of cAMP, or by direct PKA activation. We have previously shown that ryanodine pretreatment abolishes caffeineinduced CICR in clonal ␤-cells, which express functional RyRs (23). As a result, the present data do not support the idea that elevation of cAMP uncovers CICR by an action on RyRs (9,10).
Apart from the above-mentioned actions, caffeine interferes with IP 3 signaling. High concentrations thus inhibit agonistinduced formation of IP 3 (39) as well as its action on the IP 3 Rs (40), an effect observed also in pancreatic ␤-cells (51). Because of the opposite actions on RyRs and IP 3 Rs, high concentrations of caffeine have been used to discriminate between them (52). Our observation that 20 mM caffeine inhibits cAMP/PKA-promoted CICR is therefore consistent with an IP 3 R-mediated effect. This idea was further tested with the cell-permeable IP 3 R antagonist 2-APB (43). Despite its limited potency (44), 2-APB strongly inhibited Ca 2ϩ spiking promoted by glucagon and forskolin, providing additional arguments for the involvement of IP 3 Rs in CICR promoted by cAMP.
Glucose-stimulated insulin secretion depends on influx of Ca 2ϩ through voltage-dependent L-type channels. Evidence indicates a close association between these channels and the secretory granules (53). Thanks to this arrangement, the [Ca 2ϩ ] i level triggering exocytosis of the granules reaches 5-10fold higher concentrations than in the remainder of the cytoplasm. The present experiments show that [Ca 2ϩ ] i spikes caused by CICR in depolarized ␤-cells trigger a [Ca 2ϩ ] i response in neighboring reporter cells. Because this response was always delayed by at least one acquisition cycle (2 s), it probably represents release of an active messenger from the ␤-cells. This factor may be ATP, which is co-secreted with insulin (46) and is known to induce a purinergic Ca 2ϩ response in HEL cells (45). Our data consequently show that CICR is an amplifier of exocytosis in response to depolarization-dependent influx of Ca 2ϩ , perhaps indicating that CICR is acting locally to further elevate the high [Ca 2ϩ ] i levels at the site of exocytosis. The failure of some [Ca 2ϩ ] i spikes to elicit a response in the HEL cells may indicate that secretion does not always occur or that the content of active factor(s) varies between secretory granules. If secretion occurs from a ␤-cell site not facing the reporter cell, it is also possible that dilution in the medium prevents a response. Indeed, the HEL cells are apparently not sufficiently sensitive to detect secretion from the ␤-cells in the absence of CICR, although [Ca 2ϩ ] i is elevated by depolarization alone.
A recent study of mouse ␤-cells indicates the presence of an atypical CICR mechanism involving neither IP 3 Rs nor RyRs (12). This phenomenon is rather sluggish and distinctly different from the explosive Ca 2ϩ spiking characterizing the presently studied CICR. Moreover, knockout of the low-affinity Ca 2ϩ -transporting sarco(endo)plasmic reticulum ATPase 3 (SERCA-3) abolishes the atypical CICR (12) but does not affect IP 3 -mediated Ca 2ϩ -release, which depends on the high affinity SERCA-2 (54). Because SERCA-3 is activated only when [Ca 2ϩ ] i is elevated above basal levels (54), the presently used conditions with diazoxide and methoxyverapamil would prevent filling of the Ca 2ϩ pool from which the atypical CICR occurs.
In most cases, controversies regarding the type of receptor underlying CICR in pancreatic ␤-cells can be explained by the involvement of different mechanisms in clonal and primary ␤-cells (23). Nearly all published data on CICR in primary ␤-cells are consistent with a PKA-mediated effect on IP 3 Rs, even when the authors favor RyRs. The observations that the FIG. 7. CICR triggers exocytosis. HEL cells in suspension were loaded with 0.5 M fura-2/AM during incubation for 30 min. Mouse pancreatic ␤-cells attached to a coverslip were loaded for 30 min with 1 M fluo-4/AM in medium containing 11 mM glucose and 250 M diazoxide. After rinsing in the latter medium lacking indicator, HEL cells were added to the chamber with ␤-cells and covered by a permeable polycarbonate filter. The cells were then superfused with the same medium lacking indicator. As indicated by the bar, the cells were exposed to 90 mM KCl. The drawing in A illustrates the relative positions of the studied ␤-cells and HEL cells, and the traces in B show the measurements of [Ca 2ϩ ] i in each cell. The [Ca 2ϩ ] i data in ␤-cells are presented as ⌬F/F 0 ; F 0 corresponds to the average basal fluorescence before the addition of KCl and ⌬F to the deviation from F 0 . The dashed vertical lines illustrate the delay between spikes in ␤-cells and HEL cells. Of 32 Ca 2ϩ spikes in five ␤-cells, 16 were followed by spikes in eight nearby responsive HEL cells (p Ͻ 0.001).
Epac activator 8-pCPT-2Ј-O-Me-cAMP raises [Ca 2ϩ ] i and stimulates exocytosis in human ␤-cells were attributed to activation of CICR via RyRs (22). It is unfortunate that the short exposure periods (10 s) to Epac activator preclude discrimination between activation of Ca 2ϩ influx and release from a limited intracellular pool. Such discrimination is also prevented by exposure to 5.6 mM glucose, which depolarizes the ␤-cell to the threshold for opening of L-type Ca 2ϩ channels and stimulation of insulin release. Under such conditions, any minor depolarization triggers Ca 2ϩ influx. Indeed, by inhibiting the K ATP channels, cAMP depolarizes the ␤-cell and induces electrical activity even at subthreshold concentrations of glucose (55). Moreover, cAMP amplifies Ca 2ϩ influx into the ␤-cells by a direct effect on the L-type channels (56). It remains to establish whether Epac is involved in these cAMP effects on the K ATP and Ca 2ϩ channels. Although ryanodine was reported to diminish the Ca 2ϩ -elevating action of 8-pCPT-2Ј-O-Me-cAMP on the human ␤-cells (22), ryanodine has no effect on CICR elicited by Ca 2ϩ leakage from the ER in mouse and human ␤-cells (23). A stimulatory effect of Epac activation on insulin release does not require participation of CICR because there is evidence that Epac is involved in distal steps of exocytosis, explaining how cAMP amplifies Ca 2ϩ -triggered secretion (57,58).
We have demonstrated previously that typical CICR in primary pancreatic ␤-cells from mice rats and humans is caused by activation of IP 3 Rs. The present data show that this is the case as well for cAMP-promoted CICR and that the effect is mediated by PKA. Consistent with a recent suggestion that RyRs in ␤-cells may not depend on cAMP and have a different role than CICR (14), we find no evidence for their involvement in CICR even after elevation of cAMP. GLP-1 shows promising results in the treatment of type 2 diabetes (59). The present data indicate that this may be caused in part by activation of CICR, which amplifies insulin release.